0:00:00.399,0:00:09.269
♪ (preroll music) ♪
0:00:09.269,0:00:15.829
Angel: After half a year, Volkswagen committed[br]to tweaks to their emission readings.
0:00:15.829,0:00:19.939
Those two boys, Daniel Lange and[br]Felix Domke here on my left,
0:00:19.939,0:00:22.170
will share some insights with us.
0:00:22.170,0:00:24.330
Daniel will not only focus on the ECUs,
0:00:24.330,0:00:26.759
which is the acronym for the[br]Electronic Control Unit,
0:00:26.759,0:00:29.599
and I think we're seeing one[br]over here already,
0:00:29.599,0:00:34.370
whereby Felix will show us some tricks[br]to extract and tweak the firmware.
0:00:34.370,0:00:37.750
On both sides we will see how many people[br]have been involved in the entire process
0:00:37.750,0:00:42.460
and we would get an idea what everything[br]is involved in there.
0:00:42.460,0:00:47.860
So, you applause and I'm gonna take over[br]the Bildschirm.
0:00:47.860,0:00:51.440
Good luck!
0:00:53.040,0:00:56.810
Felix: Alright. Hello? Okay.
0:00:56.810,0:01:01.210
Hey, so, I'm Felix Domke.
0:01:01.210,0:01:04.220
Do we see the video output yet?[br]No.
0:01:04.220,0:01:05.740
Anyway, I'm Felix Domke.
0:01:05.740,0:01:08.470
I'm here on my own[br]because I was personally interested
0:01:08.470,0:01:12.810
in how Volkswagen is cheating[br]on their emission control.
0:01:12.810,0:01:17.450
And maybe we get video at some point.
0:01:18.890,0:01:21.150
I want to stress that I was self-funded.
0:01:21.150,0:01:26.380
I did this with my own money because I was[br]personally interested in this.
0:01:26.380,0:01:30.280
So I did not do this on behalf of anyone else.
0:01:30.280,0:01:34.869
Daniel: Let's start the Keynote again and[br]see whether that one works better then.
0:01:34.869,0:01:37.720
F: I am sure we figure this out.
0:01:37.720,0:01:40.850
D: Oh, it worked before?
0:01:42.000,0:01:47.140
Yes, that's because one of us two wanted[br]to use a Mac.
0:01:47.140,0:01:48.880
audience laughs
0:01:48.880,0:01:52.690
F: But I wanted to use Keynote,[br]I don't care which operating system.
0:01:52.690,0:01:54.080
D: This one works.
0:01:54.080,0:01:58.010
F: Anyway. So I will now hand off to Daniel[br]which will give the first part of this talk
0:01:58.010,0:02:01.020
and after that I will give[br]the second part of this talk.
0:02:01.020,0:02:04.290
D: Okay, thanks Felix.[br]My name is Daniel.
0:02:04.290,0:02:09.559
I used to work for a big Bavarian auto[br]manufacturer which is not Audi...
0:02:09.559,0:02:12.629
audience laughing
0:02:12.629,0:02:14.189
...for 14 years.
0:02:14.189,0:02:19.129
I've been running the IT strategy,[br]I've been doing IT architecture.
0:02:19.129,0:02:20.909
And most relevant to this talk,
0:02:20.909,0:02:26.200
I've been responsible for the[br]process chain electronics and electric.
0:02:26.200,0:02:30.680
I've done the rollout of[br]Connected Drive in China.
0:02:30.680,0:02:35.180
So I kind of have quite deep insight into[br]how the automotive industry works
0:02:35.180,0:02:37.219
and I'd like to share a bit with you.
0:02:37.219,0:02:41.430
I'm an Engineer by training,[br]I guess many of you are.
0:02:41.430,0:02:45.269
And I want to share[br]how the engineers think
0:02:45.269,0:02:48.799
inside such a big corporation[br]like Volkswagen,
0:02:48.799,0:02:52.560
and what pressures, what boundary[br]conditions they are working on.
0:02:52.560,0:02:55.900
I have my own company now[br]which makes my life a bit easier
0:02:55.900,0:02:59.099
than Felix's, as you see[br]in the legal disclaimer.
0:02:59.099,0:03:04.400
Those are folks from the UK.
0:03:04.400,0:03:10.150
They're called "Brandalism", I hope you[br]notice the "McDonald's"-M at the end.
0:03:10.150,0:03:15.450
Those are folks who started a few years[br]ago to reclaim the public space.
0:03:15.450,0:03:18.709
They were just annoyed by[br]all of that advertising.
0:03:18.709,0:03:24.049
And when the Paris negotiations[br]for the eco treatment came,
0:03:24.049,0:03:27.790
they just felt a big invitation to[br]use the opportunity
0:03:27.790,0:03:30.650
that Volkswagen has created[br]for all of us
0:03:30.650,0:03:34.639
and make advertising in their style, but[br]perhaps not in the message
0:03:34.639,0:03:36.879
they would usually have conveyed.
0:03:36.879,0:03:40.749
I'm a strategist.
0:03:40.749,0:03:46.939
So what is the thing that defines how the[br]automotive indusry works today?
0:03:46.939,0:03:50.709
We are in a saturated market.
0:03:50.709,0:03:56.579
In the developed countries, so everywhere[br]in Europe, in the North Americas,
0:03:56.579,0:03:59.639
everybody has a car that wants one.[br]Some have two.
0:03:59.639,0:04:01.870
So when you want to sell another car
0:04:01.870,0:04:07.049
you're basically talking about replacing[br]an existing car with another one.
0:04:07.049,0:04:09.779
The only growth you have[br]is in the BRIC states.
0:04:09.779,0:04:12.969
BRIC is: Brazil, Russia, India, and China.
0:04:12.969,0:04:15.989
And, here, especially in China.
0:04:15.989,0:04:19.978
You have a big overcapacity. There's just[br]too many automotive manufacturers
0:04:19.978,0:04:22.419
and there's too many plants they have.
0:04:22.419,0:04:26.690
So all of them basically struggle[br]to get the loads on the plants,
0:04:26.690,0:04:31.729
to produce enough cars and to have[br]those cars sold at some point in time.
0:04:31.729,0:04:35.050
Because the queueing in between[br]production and sales
0:04:35.050,0:04:41.849
is actually the big parking spaces you see[br]in Bremerhaven or so
0:04:41.849,0:04:45.979
where there's ten thousands, in some[br]countries even hundreds of thousands of cars
0:04:45.979,0:04:50.800
basically stored in between[br]production and sales.
0:04:50.800,0:04:53.629
Ten years ago, fifteen years ago,[br]that didn't exist.
0:04:53.629,0:04:58.419
The cars were basically sold off the factory.
0:04:58.419,0:05:02.770
But people have been moving away[br]very, very slowly from cars.
0:05:02.770,0:05:04.539
They have, as I said, a saturated market.
0:05:04.539,0:05:08.819
It's just not that easy[br]to sell a car anymore.
0:05:08.819,0:05:12.259
That is because of social shifts.
0:05:12.259,0:05:15.620
When I was young, there was[br]"The Dukes of Hazzard",
0:05:15.620,0:05:20.289
there was "General Lee", this car[br]that basically is the star of the show.
0:05:20.289,0:05:24.500
There was "Knight Rider" and nobody watched[br]it for David Hasselhoff, not even the girls.
0:05:24.500,0:05:27.870
They watched it for KITT.
0:05:27.870,0:05:30.720
When I was young, I wanted to own a car.
0:05:30.720,0:05:33.810
I wanted to have KITT, possibly.
0:05:33.810,0:05:37.430
And when I grew old enough, I found out[br]I can get a car that looks like KITT
0:05:37.430,0:05:39.710
but, you know, all the[br]fun stuff is not in there,
0:05:39.710,0:05:43.840
so I turned to computers.
0:05:43.840,0:05:48.710
The next thing is organization,[br]megacities.
0:05:48.710,0:05:51.259
We live in very condensed spaces[br]in those cities.
0:05:51.259,0:05:56.460
If you talk about a place like Beijing[br]where there's like 21 million people
0:05:56.460,0:06:01.669
in an area that is one city,[br]where there's nothing big inbetween,
0:06:01.669,0:06:04.259
there's no river, there's no forest,
0:06:04.259,0:06:06.389
it's just like one city.
0:06:06.389,0:06:09.990
If you go to Tokyo, Yokohama,[br]you can drive on the motorway
0:06:09.990,0:06:14.879
for nearly three hours when you enter[br]the city before you leave the city.
0:06:14.879,0:06:18.229
And you're driving on the motorway,[br]you're driving on an elevated road
0:06:18.229,0:06:21.620
for which you paid toll[br]so you actually can drive.
0:06:21.620,0:06:25.099
But it's three hours before[br]you leave the city again.
0:06:25.099,0:06:28.050
And in these cities owning a car[br]and operating a car
0:06:28.050,0:06:31.710
is about the worst thing you can do.[br]You just don't want to do that.
0:06:31.710,0:06:41.039
The average speed of a car in Beijing[br]these days is 12 km/h.
0:06:41.039,0:06:43.819
If you're a good runner you can beat that.
0:06:43.819,0:06:49.180
And incidentally this is exactly the speed[br]that a horse carriage makes.
0:06:49.180,0:06:55.889
audience laughs and applauds
0:06:55.889,0:07:01.039
We have managed to undo[br]all of the innovation of the last 200 years,
0:07:01.039,0:07:05.729
it's just the interior[br]is a little bit more comfortable.
0:07:05.729,0:07:10.189
The actual getting from A to B is the same[br]as with a horse carriage these days.
0:07:10.189,0:07:12.330
And then there's technology shifts.
0:07:12.330,0:07:17.879
The problem is, there are big things,[br]big visions that everybody follows,
0:07:17.879,0:07:20.919
like electric mobility.[br]Electric mobility means:
0:07:20.919,0:07:24.449
You buy a car that's one and a half times[br]the price of your standard car,
0:07:24.449,0:07:28.870
you lug around 300 kg of batteries for[br]no apparent reason to do so,
0:07:28.870,0:07:31.789
and you now need to install something[br]in your garage —
0:07:31.789,0:07:34.780
which you most probably don't have,[br]look at megacities —
0:07:34.780,0:07:38.740
to be able to recharge the car[br]because it only goes a hundred miles.
0:07:38.740,0:07:44.960
So it's currently not a very compelling[br]thing to sell to the end customer.
0:07:44.960,0:07:49.430
There's self-driving cars, which is kind of[br]a great, big vision.
0:07:49.430,0:07:51.520
But I would really call that a "vision".
0:07:51.520,0:07:56.169
A "vision" is something that's not being[br]implemented in my lifetime.
0:07:56.169,0:07:57.569
And then there's downsizing.
0:07:57.569,0:07:59.689
Downsizing means ...
0:07:59.689,0:08:01.699
Everybody wanted to have[br]the biggest engine,
0:08:01.699,0:08:04.770
everybody wanted to have the[br]biggest car, let's say, 10 years ago.
0:08:04.770,0:08:08.580
You wanted to have that six cylinder[br]that was giving you status.
0:08:08.580,0:08:12.289
But now the automotive industry[br]has an overall cap
0:08:12.289,0:08:17.469
on how much emissions the average[br]new car fleet my have.
0:08:17.469,0:08:22.009
And they can only reach that if they manage[br]to sell smaller engines to you.
0:08:22.009,0:08:24.349
Because for everybody who buys[br]a really big engine
0:08:24.349,0:08:26.999
that will never ever make that emission cap,
0:08:26.999,0:08:30.240
they need somebody whom they've[br]sold a small car to —
0:08:30.240,0:08:34.909
preferrably an electric car, because[br]they even have statistical advantages
0:08:34.909,0:08:37.220
to make them a bit more attractive —
0:08:37.220,0:08:38.820
to set that off.
0:08:38.820,0:08:43.230
So very literally the poor guy[br]with the small car needs to exist
0:08:43.230,0:08:47.850
for the rich guy that drives the[br]eight cylinder and doesn't give a shit.
0:08:50.700,0:08:57.940
Strategy-wise, there's only two things[br]that an automotive car company is driven in.
0:08:57.940,0:09:00.980
And that's really everything there is.
0:09:00.980,0:09:03.340
There is "reach a target ROCE".
0:09:03.340,0:09:07.260
ROCE is: Return on Capital Employed.[br]That is just two numbers:
0:09:07.260,0:09:12.200
your EBIT, which is your Earnings[br]Before Income Tax,
0:09:12.200,0:09:17.230
and the amount of money you have[br]in your company, the employed capital,
0:09:17.230,0:09:19.160
which you got from people[br]that lent it to you
0:09:19.160,0:09:22.660
or from your stakeholders, from your investors.
0:09:22.660,0:09:25.520
And that is what the company[br]is measured against.
0:09:25.520,0:09:27.900
Every automotive company[br]basically runs like this.
0:09:27.900,0:09:30.870
Just this one figure,[br]it's a percentage like "30%."
0:09:30.870,0:09:39.330
"30%" means: On the money you have[br]you made a 30% return during that year.
0:09:39.330,0:09:44.750
The downside of measuring in ROCE is that[br]everytime you use that Euro or Dollar
0:09:44.750,0:09:48.520
it counts again because[br]the money works for you.
0:09:48.520,0:09:50.010
That means what you're looking at
0:09:50.010,0:09:55.480
is a company that gradually moves[br]from a very industrial type of application
0:09:55.480,0:09:57.370
to something that tries to move faster,
0:09:57.370,0:10:03.180
that tries to be quick and[br]regain money faster.
0:10:03.180,0:10:05.100
And then there's[br]"outperform the competition."
0:10:05.100,0:10:08.200
You have to understand the situation
0:10:08.200,0:10:10.790
that there's a good dozen companies
0:10:10.790,0:10:14.420
and everybody has the[br]same strategic position:
0:10:14.420,0:10:17.240
"We will outperform the competition."
0:10:17.240,0:10:20.140
So statistically, you will know that[br]half of them are going to fail
0:10:20.140,0:10:25.460
because that won't happen, right?[br]Somebody has to be the lower half.
0:10:25.460,0:10:29.220
But the only thing I have seen[br]in about five or six companies
0:10:29.220,0:10:33.180
where I know the strategy in detail,[br]is: the sequence.
0:10:33.180,0:10:36.680
Is the first or is the latter[br]the more important one?
0:10:36.680,0:10:38.710
And sometimes that depends on markets.
0:10:38.710,0:10:41.890
There's this new emerging market and[br]you want to outperform the competition,
0:10:41.890,0:10:43.620
you want to grow more.
0:10:43.620,0:10:48.180
And then there's this laggard market somewhere[br]in the European Union
0:10:48.180,0:10:51.870
where you just look at the money, you know,[br]how much money are we making on this.
0:10:51.870,0:10:56.810
But that's all, that is how an engineer is[br]basically steered, that is the strategy.
0:10:56.810,0:11:03.530
And that means when you break that down[br]through the levels of hierarchy, what is counting is:
0:11:03.530,0:11:07.330
How much money do you[br]need to make this?
0:11:07.330,0:11:09.720
How much money are you[br]gonna make on this?
0:11:09.720,0:11:14.080
Those two divided will be[br]contributing to the ROCE.
0:11:14.080,0:11:20.320
And do you deliver anything that can[br]help us outperform the competition?
0:11:20.320,0:11:24.380
You notice that there is a lack, which is,[br]you know: What does the customer want?
0:11:24.380,0:11:27.790
Or: What is good for associates?[br]Or something like that.
0:11:27.790,0:11:30.720
Just in case you hadn't noticed before.
0:11:30.720,0:11:37.460
Okay, I'd like to do a bit of a quiz with[br]you before you all fall asleep after lunch.
0:11:37.460,0:11:39.270
Eleven million.
0:11:39.270,0:11:42.800
"Eleven million" in the context of the[br]exhaust emission scandal.
0:11:42.800,0:11:44.320
What is that number?
0:11:44.320,0:11:46.890
Audience: Cars affected!
0:11:46.890,0:11:48.830
Correct! Cars affected.
0:11:48.830,0:11:51.960
Eleven million is actually the[br]Volkswagen cars
0:11:51.960,0:11:54.910
which need to be recalled world-wide
0:11:54.910,0:12:00.050
to get this little filter thing fixed[br]and their software updated
0:12:00.050,0:12:04.880
to meet the emission targets which they[br]had been produced against.
0:12:04.880,0:12:07.580
1500 ...?
0:12:09.810,0:12:12.380
A: Number of engineers!
0:12:12.380,0:12:14.470
Number of engineers?[br]No, not correct.
0:12:14.470,0:12:20.420
Number of engineers would be above 10000[br]for a car in Volkswagen Group. Sorry?
0:12:20.420,0:12:22.880
A: Cost for fixing it per car?
0:12:22.880,0:12:26.750
Cost for fixing it per car? No, that's[br]maximum 600, we're gonna see later.
0:12:26.750,0:12:28.580
unintelligible suggestion from audience
0:12:28.580,0:12:32.100
No? Well that was too difficult then,[br]and that was a bit intentional.
0:12:32.100,0:12:37.080
That's the amount of hard disks they[br]collected from the associates.
0:12:37.080,0:12:42.420
audience laughs and applauds
0:12:42.420,0:12:48.190
Now the thing is, we've had in the press[br]that there is maximum 13 managers
0:12:48.190,0:12:53.740
which are responsible for this[br]emission scandal within Volkswagen.
0:12:53.740,0:12:59.780
But then they collect 1500 hard disks and[br]USB sticks from 380 associates,
0:12:59.780,0:13:03.670
and that number is a month old because[br]they haven't reported newer numbers.
0:13:03.670,0:13:07.680
So something is mismatching there, right?[br]Something is mismatching there.
0:13:07.680,0:13:09.980
So the first number we have is[br]for how many associates
0:13:09.980,0:13:13.190
are actually somehow[br]affected by this is 380.
0:13:13.190,0:13:17.780
Because you come to work somewhere in[br]Wolfsburg I think, right?
0:13:17.780,0:13:21.230
And then there's this nice chap coming up[br]and telling you,
0:13:21.230,0:13:24.170
"Uhm, actually we took[br]the hard disk off your PC,
0:13:24.170,0:13:26.880
you're gonna get a new one from IT,[br]we guess tomorrow,
0:13:26.880,0:13:31.920
they're a bit behind with, you know ..."
0:13:31.920,0:13:35.190
6.7 billion ...?
0:13:35.190,0:13:37.260
Just shout!
0:13:37.260,0:13:39.400
unintelligible suggestions from audience
0:13:39.400,0:13:42.640
Fine? No that will be less, much less.
0:13:42.640,0:13:44.050
unintelligible suggestions from audience
0:13:44.050,0:13:46.880
Yes, you're getting close.[br]It's the money they put back,
0:13:46.880,0:13:52.220
they set aside to actually pay[br]for the recall and the legal fees.
0:13:52.220,0:13:57.330
Now if you divide that by 11 million[br]you get about €600 per car.
0:13:57.330,0:14:01.760
So it's not that much money per car.
0:14:01.760,0:14:08.040
In Europe, the plan is basically that you[br]go to the dealer and get a software update.
0:14:08.040,0:14:14.980
In the States, people already got $1000[br]in cash and in coupons
0:14:14.980,0:14:17.260
as a goodwill measure.
0:14:17.260,0:14:22.770
So something I learnt from Martin Haase[br]here going to the CCC Congress all the time
0:14:22.770,0:14:26.150
is that we need to read text really well.
0:14:26.150,0:14:30.350
So the upper one is the original in German,[br]the lower one is my English translation.
0:14:30.350,0:14:34.870
The English translation is as accurate as[br]possible, so it's not good English.
0:14:34.870,0:14:40.070
Please excuse that, it is so you get the gist[br]in case you can only read the English.
0:14:40.070,0:14:44.400
So that is Mr. Pötsch, he's the president[br]of the Volkswagen supervisory board.
0:14:44.400,0:14:50.170
He is the poor guy that now[br]has to sort it all out.
0:14:50.170,0:14:55.290
He used to be the CFO. We're gonna see why[br]that is important a little bit later.
0:14:55.290,0:14:58.740
And he has made this analysis:
0:14:58.740,0:15:03.870
It was "individual misbehaviour",[br]so it's not an organizational problem,
0:15:03.870,0:15:06.630
it's "weaknesses in particular processes",
0:15:06.630,0:15:10.250
and it's "the attitude in[br]particular sub-partitions" ...
0:15:10.250,0:15:14.060
"Teilbereiche des Unternehmens",[br]it's impossible to translate in English,
0:15:14.060,0:15:19.850
it's actually impossible in German, but,[br]you know, the legal team came up with that.
0:15:19.850,0:15:24.620
So the "attitude in particular sub-partitions[br]of the company to tolerate rule violations."
0:15:24.620,0:15:28.590
Now, if we go through this very quickly:[br]It's not a rule violation,
0:15:28.590,0:15:32.050
you violated the fucking law.
0:15:32.050,0:15:37.160
The other thing is, if you have particular[br]processes, you have particular associates,
0:15:37.160,0:15:40.010
and you have particular sub-partitions[br]of the company,
0:15:40.010,0:15:43.560
That tells you something, right?[br]That just tells you something.
0:15:43.560,0:15:46.610
This was probably two days' work of[br]somebody in the legal team,
0:15:46.610,0:15:51.030
and I guess you noticed, right?[br]I guess you notice.
0:15:51.030,0:15:55.680
"Legal team" is probably these people.
0:15:55.680,0:16:02.040
Jones Day is a big American lawyer company[br]
0:16:02.040,0:16:05.320
and they've asked them to help[br]with sorting out this.
0:16:05.320,0:16:11.530
Now the funny thing is, there's public prosecutors[br]all over the planet interested in Volkswagen
0:16:11.530,0:16:13.980
but Volkswagen thinks it's[br]not really clever
0:16:13.980,0:16:17.320
to have those people come in[br]and find all the info,
0:16:17.320,0:16:22.020
it's better to have Jones Day,[br]their own kind of bought-in legal team,
0:16:22.020,0:16:24.690
ask the associates first.
0:16:24.690,0:16:31.040
Now the problem is, whenever the let's say[br]German prosecutors wake up and go in there
0:16:31.040,0:16:36.120
and say, like, "We would like[br]to see what has happened,
0:16:36.120,0:16:39.530
so please hand over the material,[br]please hand over the hard disks,"
0:16:39.530,0:16:44.630
they would get a very, very nice reception,[br]be greeted with coffee and shown a room
0:16:44.630,0:16:47.100
where all of the hard disks[br]and everything is stored,
0:16:47.100,0:16:51.990
"We collected it for you."
0:16:51.990,0:16:55.860
I have no idea whether they gonna show[br]everything to them
0:16:55.860,0:17:00.890
I have no idea whether there may be[br]some material lost in between.
0:17:00.890,0:17:04.680
We've heard from Anna earlier[br]in Germany it seems to be
0:17:04.680,0:17:08.339
that things hit the shredder and[br]hard disks get lost and everything.
0:17:08.339,0:17:13.010
So if it works like that in the government[br]I have no idea how it works in companies.
0:17:13.010,0:17:17.540
But if I was on the prosecutors I'd[br]probably see that I speed up a little
0:17:17.540,0:17:20.540
because otherwise you'll get[br]all pre-prepared material.
0:17:20.540,0:17:23.010
And because Jones Day[br]can't do all of that —
0:17:23.010,0:17:26.310
you have to interview all of those people,[br]and you have to look through the hard disks —
0:17:26.310,0:17:29.450
they asked Deloitte[br]to come in and help them.
0:17:29.450,0:17:32.810
Now Deloitte are a very good company,[br]they have very, very good forensic teams,
0:17:32.810,0:17:36.020
so that's a very good choice.[br]But the important thing here is:
0:17:36.020,0:17:41.160
Out of the four big consulting companies[br]that do finance analysis and stuff
0:17:41.160,0:17:45.420
those are the only Americans. The others[br]are headquartered somewhere else.
0:17:45.420,0:17:52.080
So what it tells you here —[br]American legal teams, American auditors —
0:17:52.080,0:17:55.940
that's where Volkswagen looks.[br]Volkswagen is actually afraid of America.
0:17:55.940,0:18:02.510
They are not that afraid of Europe or some[br]other country in some other continent.
0:18:05.110,0:18:08.990
Now, let's talk text a bit again.
0:18:10.910,0:18:16.750
"We have no findings on the involvement[br]of the supervisory board
0:18:16.750,0:18:19.630
or the board of management presented."
0:18:19.630,0:18:24.590
Now, again, "no findings", okay,[br]"presented", right?
0:18:24.590,0:18:28.200
It's not "we don't have any findings"[br]or "there is nothing",
0:18:28.200,0:18:30.810
it says "we have no findings presented."
0:18:30.810,0:18:34.040
And the other thing is "involvement",[br]that's an odd term.
0:18:34.040,0:18:36.670
In German, "Involvierung",[br]that's not even German, right?
0:18:36.670,0:18:40.610
If you look it up, "Involvierung", nobody[br]of you talks of "Involvierung"
0:18:40.610,0:18:44.260
when you talk to your family or[br]when you do something at work.
0:18:44.260,0:18:49.750
The trick here is, the supervisory board[br]has a reason for existing: supervision.
0:18:49.750,0:18:54.480
audience laughs and applauds
0:18:54.480,0:18:59.140
The board of management has a reason for[br]existing and that is: decision.
0:18:59.140,0:19:01.050
They are the deciding body.
0:19:01.050,0:19:04.500
None of them are ever "involved", right?
0:19:04.500,0:19:07.740
When you work on something[br]in a big hierarchical company
0:19:07.740,0:19:10.840
there is no "involvement"[br]of your board member,
0:19:10.840,0:19:14.380
there is no "involvement"[br]of your supervisory board member.
0:19:14.380,0:19:18.120
So per definition, they cannot[br]have an involvement, right?
0:19:18.120,0:19:21.290
If he wanted to be straight[br]he would have said:
0:19:21.290,0:19:27.250
"I, as a former board of management director[br]and now as the head of the supervisory board,
0:19:27.250,0:19:32.520
I guarantee there was no involvement[br]of my or my colleagues in this.
0:19:32.520,0:19:37.370
And if there was, I would pay back my[br]salary, I will go to jail, I will ... whatever."
0:19:37.370,0:19:40.320
Right, sacrifice a goat?
0:19:40.320,0:19:42.970
But that would have been[br]straight communication.
0:19:42.970,0:19:47.250
But this is not straight communication,[br]this is... bullshit.
0:19:48.430,0:19:52.020
Okay, quiz time![br]10 ...?
0:19:52.020,0:19:59.070
You remember, this guy here told us there's[br]no involvement in anything fishy, right?
0:19:59.070,0:20:03.290
It's all those small engineers,[br]all those bad, bad people down there.
0:20:03.290,0:20:05.460
But they are gonna hunt 'em down, right?
0:20:05.460,0:20:08.110
So there's no involvement[br]with anything fishy here.
0:20:08.110,0:20:10.530
So, in that context, what is "10"?
0:20:10.530,0:20:12.140
A: Board members!
0:20:12.140,0:20:14.190
10 board members?[br]Close, they have a little more.
0:20:14.190,0:20:15.990
A: Levels of hierarchy.
0:20:15.990,0:20:20.120
Levels of hierarchy — quite good. It's,[br]I think, eight or so, but you're quite close.
0:20:20.120,0:20:29.120
No, it's actually the amount of planes[br]that Volkswagen owns.
0:20:29.120,0:20:30.930
All of them are jet planes.
0:20:30.930,0:20:35.260
Because if you're a board member[br]you have to, you know, fly in style.
0:20:35.260,0:20:38.740
And because there's nothing[br]ever fishy at Volkswagen
0:20:38.740,0:20:43.670
it's run by Lion Air Services[br]out of the Braunschweig airport.
0:20:43.670,0:20:49.870
And obviously, Lion Air Servives is registered[br]in Georgetown on the Cayman Islands.
0:20:49.870,0:20:52.870
applause and laughter
0:20:52.870,0:20:54.930
Nothing fishy ever in that company.
0:20:54.930,0:20:58.190
Okay, let's get back to topic. I have[br]about another ten minutes
0:20:58.190,0:21:03.450
before I want to get Felix the chance to[br]show you what he has done on the ECUs.
0:21:03.450,0:21:09.940
So I need to get you up to speed[br]about how all of this context here works.
0:21:09.940,0:21:15.930
And this here is called the NEDC,[br]it's the New European Driving Cycle.
0:21:15.930,0:21:20.860
This is what your car is tested[br]against for emissions.
0:21:20.860,0:21:24.620
It works like that: You condition[br]the vehicle a day before.
0:21:24.620,0:21:27.340
Which means you really[br]drive it hard on the Autobahn
0:21:27.340,0:21:31.370
so the exhaust is really free[br]and everything.
0:21:31.370,0:21:37.490
And then you do these cycles here where[br]you basically accelerate the vehicle,
0:21:37.490,0:21:40.200
slow down, accelerate the vehicle,[br]slow down, accelerate the vehicle,
0:21:40.200,0:21:43.680
slow a bit down, slow a bit more down,[br]and then you cycle again.
0:21:43.680,0:21:47.090
And the last, cycle 5, is an optional one,[br]depending on what you measure,
0:21:47.090,0:21:49.390
that is actually going to the autobahn
0:21:49.390,0:21:56.280
and you're going up to a top speed of[br]120 km/h for a very short period of time.
0:21:56.280,0:22:04.850
The people that have detected the[br]tweaked emissions
0:22:04.850,0:22:11.230
in the VW Jetta and Passat they looked at,[br]they have called this
0:22:11.230,0:22:16.450
"a very light usage cycle,"[br]and they called it "unrealistic".
0:22:16.450,0:22:21.260
Because basically nobody drives the car[br]like this, it's a very artificial thing.
0:22:21.260,0:22:23.210
And that is the problem[br]for the engineer, right?
0:22:23.210,0:22:27.630
The engineer looks at this and says,[br]"Yeah, you know, it's a standard.
0:22:27.630,0:22:30.010
It's something we do to measure against."
0:22:30.010,0:22:32.930
But nobody drives like this.[br]It's not realistic, right?
0:22:32.930,0:22:37.680
So if you fake the data in this we're not[br]actually faking something our customer uses
0:22:37.680,0:22:41.800
because no customer drives like this,[br]it's very artificial.
0:22:41.800,0:22:46.200
And there's a very good report by ICCT[br]which is, "Mind the Gap".
0:22:46.200,0:22:49.250
Which is what you hear in London when you[br]go into the Tube.
0:22:49.250,0:22:55.059
And what they mean is the gap between what[br]gets out when you measure emissions like this
0:22:55.059,0:22:57.850
and what gets out when you[br]actually drive the car.
0:22:57.850,0:23:00.870
And that gap is widening[br]year by year by year.
0:23:00.870,0:23:05.740
Because engineers get better and better[br]at optimizing for this cycle.
0:23:05.740,0:23:10.770
The cars on the street? Phhh, they do get[br]better as well, but less, right?
0:23:10.770,0:23:14.140
That's why the gap widens.
0:23:14.140,0:23:17.420
And trickery on those tests[br]is very common.
0:23:17.420,0:23:21.130
I'm sorry you can't probably[br]read that in the stream
0:23:21.130,0:23:23.800
and probably can't read that[br]when you're back down there.
0:23:23.800,0:23:27.010
But that's an original slide I had to take[br]from Transport & Environment,
0:23:27.010,0:23:29.180
from that report which I just named.
0:23:29.180,0:23:32.660
And what it says there is[br]what tricks people are doing
0:23:32.660,0:23:35.320
to actually drive down the emissions.
0:23:35.320,0:23:37.900
For example, they blow up the tyres
0:23:37.900,0:23:41.770
by 3 bars more than you could[br]actually use them on the road.
0:23:41.770,0:23:45.600
Now when you do, the bottom of the tyre[br]looks like this, right?
0:23:45.600,0:23:48.340
So that means you only have a very, very[br]small portion of the tyre
0:23:48.340,0:23:52.540
that still touches the ground,[br]so your resistance gets reduced.
0:23:52.540,0:23:59.760
They put diesel into the oil beause diesel[br]is lighter than the oil which you are using
0:23:59.760,0:24:02.740
inside the vehicle, so friction gets reduced.
0:24:02.740,0:24:07.320
They take off the mirror, the side mirror[br]on the passenger side
0:24:07.320,0:24:10.290
because that is not legally[br]required to be existing,
0:24:10.290,0:24:14.170
so, you know, it's resistance,[br]so get away with it.
0:24:14.170,0:24:17.790
They tape close all of the[br]openings of the vehicle
0:24:17.790,0:24:19.980
because obviously when the[br]wind goes over it
0:24:19.980,0:24:23.320
it goes much smoother[br]once you have everything taped.
0:24:23.320,0:24:30.220
Now all of these things are either okay or[br]they are kind of borderline grey area.
0:24:30.220,0:24:33.179
And they do this. This is how[br]actually emissions are tested.
0:24:33.179,0:24:36.170
So this is why an engineer,[br]when he looks at this, says,
0:24:36.170,0:24:40.160
"Yeah, it's an optimization problem. They[br]want me to get a low number
0:24:40.160,0:24:45.780
and I have pretty clever ideas, which involve[br]diesel and sticky tape and everything,
0:24:45.780,0:24:47.460
to reduce the number."
0:24:47.460,0:24:50.460
sighs
0:24:50.460,0:24:53.140
The results are this.
0:24:53.140,0:24:58.670
That's from a 2012 report from — a 2013[br]report, I'm sorry — from ADAC,
0:24:58.670,0:25:02.860
the German MRT company.
0:25:02.860,0:25:07.370
And what you see, the lighter blue ones[br]are actually the emissions
0:25:07.370,0:25:11.679
which the car produces in this cycle.
0:25:11.679,0:25:14.170
The darker blue ones are the ones[br]which are produced
0:25:14.170,0:25:17.490
when you just go on the motorway[br]and drive them.
0:25:17.490,0:25:19.140
And you see that there is a discrepancy
0:25:19.140,0:25:26.960
which is ten times, twenty times, thirty[br]times what is the measured data.
0:25:26.960,0:25:31.610
So what you need to understand is that[br]even in the past nobody ever thought,
0:25:31.610,0:25:35.980
nobody in the industry ever thought that[br]the data which was measured
0:25:35.980,0:25:40.160
had any real connection with reality, right?
0:25:40.160,0:25:43.510
The only connection was, you knew that[br]what you're measuring
0:25:43.510,0:25:48.090
within the duty cycle NEDC[br]is definitely less
0:25:48.090,0:25:51.080
than what you would ever see[br]in any realtime use.
0:25:51.080,0:25:54.410
But that's it, that's it.[br]That's no secret, right?
0:25:54.410,0:25:58.470
It's something that has been[br]out there for years.
0:25:59.630,0:26:04.690
Now the folks at Deutsche Umwelthilfe,[br]which are actually people that helped
0:26:04.690,0:26:09.130
find out what Volkswagen did, they wanted[br]to see that others do it as well.
0:26:09.130,0:26:12.460
And because I wanted to give you as much[br]information as possible
0:26:12.460,0:26:17.910
we are going to look at this product here now,[br]which is not a Volkswagen as you may see.
0:26:17.910,0:26:21.640
And when you measure this car[br]it actually looks like this.
0:26:21.640,0:26:26.809
So that means when the car is thinking it[br]is running an NEDC —
0:26:26.809,0:26:30.880
because it is conditioned to do so,[br]it is the right temperature,
0:26:30.880,0:26:35.000
it is the right setup —[br]it actually delivers the blue bars.
0:26:35.000,0:26:39.760
And if you run it because you just run it[br]and you don't do the conditioning
0:26:39.760,0:26:42.070
it delivers the grey bars.
0:26:42.070,0:26:45.610
Now there's many things you can say[br]about how they measured this
0:26:45.610,0:26:52.020
because, obviously, this is not science to[br]the best level of accuracy.
0:26:52.020,0:26:54.980
But you do see a pattern here,[br]and you do see the pattern
0:26:54.980,0:26:59.610
of the 30-, 35-fold emissions.[br]And that is what you always see
0:26:59.610,0:27:03.170
because this is what an engine[br]like the one in this car —
0:27:03.170,0:27:06.370
a 1.6 l diesel engine[br]if I remember correctly —
0:27:06.370,0:27:10.600
actually does when it's just[br]operated normally.
0:27:10.600,0:27:16.800
And the lower ones are the ones which you[br]get when the engineers did all the good tweaking.
0:27:18.420,0:27:23.660
Now why has all of this ...[br]Oh sorry, so this is just one test, right?
0:27:23.660,0:27:28.540
And you see that this test,[br]when the vehicle is cold,
0:27:28.540,0:27:34.799
you get fresh air with a nice rose smell[br]out of the exhaust.
0:27:34.799,0:27:38.660
And when the vehicle is operated normally[br]you basically get what you expect,
0:27:38.660,0:27:42.309
you get the combustion products[br]out of burning diesel.
0:27:42.309,0:27:44.929
Now why is all of this now a problem?
0:27:44.929,0:27:48.110
This is now a problem because of the[br]American legal system.
0:27:48.110,0:27:50.380
The American legal system is[br]very, very different
0:27:50.380,0:27:55.370
from what people in the European Union[br]are used to.
0:27:55.370,0:27:59.510
In America, there are two things which are[br]a bit strange perhaps
0:27:59.510,0:28:03.570
to somebody who's accustomed with a[br]German legal system.
0:28:03.570,0:28:05.580
The first thing is, there's jurys.
0:28:05.580,0:28:10.490
So there's common people that actually[br]decide about what's right or what's wrong.
0:28:10.490,0:28:14.260
And that means, what they[br]award as compensation
0:28:14.260,0:28:20.390
to people that have had a disadvantage[br]are often astronomic figures.
0:28:20.390,0:28:23.690
Now these figures are sometimes[br]reduced again by the judges,
0:28:23.690,0:28:30.080
but it's not uncommon that if something[br]hurt you or you got into an accident
0:28:30.080,0:28:32.620
you're awarded million dollar sums.
0:28:32.620,0:28:37.150
In Germany, if somebody shoots your[br]eye out, you may be getting €100,000.
0:28:37.150,0:28:39.790
So there's a huge discrepancy there.
0:28:39.790,0:28:43.290
And the other thing is, in America[br]there are punitive damages.
0:28:43.290,0:28:47.830
"Punitive damages" means: You did[br]something wrong, you did it on purpose,
0:28:47.830,0:28:50.420
and you're punished for it.
0:28:50.420,0:28:54.200
In Europe, a company basically is,[br]you did something wrong
0:28:54.200,0:28:58.580
so now you have to compensate the[br]disadvantage somebody else had.
0:28:58.580,0:29:03.400
So to a certain extent,[br]a company that doesn't try to trick
0:29:03.400,0:29:09.679
actually kind of loses an opportunity because[br]if they are not detected to be tricking
0:29:09.679,0:29:11.320
they have just saved money.
0:29:11.320,0:29:15.360
There's no punitive element, there's no[br]"You will go to jail for this."
0:29:15.360,0:29:19.950
At least in this context of[br]environmental regulation.
0:29:19.950,0:29:23.799
Now in case you couldn't read that, that's[br]actually a sign I took in california.
0:29:23.799,0:29:27.850
You go into a store and it tells you[br]that basically everything you see there
0:29:27.850,0:29:33.780
and touch there is giving you cancer and[br]your unborn children will be damaged.
0:29:33.780,0:29:36.240
This is what it says there: Belts, shoes,[br]jewellery, handbags,
0:29:36.240,0:29:38.030
all products with metal, and everything
0:29:38.030,0:29:41.929
causes cancer, birth defects,[br]and other reproducive damages.
0:29:41.929,0:29:43.809
So this is America, right?
0:29:43.809,0:29:51.030
Their view of protecting the consumer is[br]completely different from Europe.
0:29:51.030,0:29:54.670
And this is why Volkswagen goes and says,[br]"We will show good faith.
0:29:54.670,0:30:00.130
We will give you, American Volkswagen[br]owner, a thousand dollars
0:30:00.130,0:30:05.340
because we just wanna make sure that[br]you at least know we care."
0:30:05.340,0:30:07.790
It's important that you care because[br]the jury will say,
0:30:07.790,0:30:11.160
"Well at least they awarded $1000,[br]maybe a little too little,
0:30:11.160,0:30:13.100
but at least they did something."
0:30:13.100,0:30:18.210
The jury would say that. A professional[br]judge in Germany would say, "Pshh, why?"
0:30:18.210,0:30:22.880
So this is why as a European customer[br]you actually go to the dealership,
0:30:22.880,0:30:25.490
and if that guy is really nice[br]you may be getting a coffee
0:30:25.490,0:30:28.660
while you wait the hour[br]that he flashes your car.
0:30:28.660,0:30:34.160
So that's the only thing you're currently[br]supposedly getting in Europe.
0:30:34.160,0:30:38.799
Okay, now the problem is:[br]What they did hurts.
0:30:38.799,0:30:42.230
And it hurts because,[br]if you do the statistics ...
0:30:42.230,0:30:49.820
Very nice people have published a[br]publication here, a real scientific publication
0:30:49.820,0:30:52.380
where they did the maths,[br]and they say:
0:30:52.380,0:30:55.830
59 people may be dying[br]earlier in the United States
0:30:55.830,0:30:59.480
because of the additional emissions[br]in the environment
0:30:59.480,0:31:02.530
which they took in and which may[br]damage their body.
0:31:02.530,0:31:06.090
The social cost of treating those people —[br]because they may be developing cancer,
0:31:06.090,0:31:08.110
they may be going to a hospital,[br]and so on —
0:31:08.110,0:31:12.100
is about 450 million Euros.[br]Now that's statistics, right?
0:31:12.100,0:31:17.429
"Lies, damn lies, and statistics."[br]Mark Twain is often quoted with that.
0:31:17.429,0:31:21.210
But the problem is: That is a real cost,[br]it is a real damage.
0:31:21.210,0:31:27.960
If you do violate emission laws it is[br]something that is damaging people's health.
0:31:27.960,0:31:31.270
It may be something that is difficult[br]to prove statistically,
0:31:31.270,0:31:36.360
but it is something which you don't only[br]do to save money here or there,
0:31:36.360,0:31:39.240
it is something which you do[br]to actually hurt people.
0:31:40.670,0:31:46.880
Okay, I need to speed up a bit. Very[br]sorry, skip this, that's the next quiz.
0:31:46.880,0:31:51.020
15.9 million is actually the salary[br]of this guy here.
0:31:53.030,0:31:57.220
That's a lady from BMW,[br]I just wanted to put that out there.
0:31:57.220,0:31:59.919
She says, "It shouldn't be called[br]Dieselgate, it's Volkswagen-Gate.
0:31:59.919,0:32:02.540
We never did anything wrong at BMW."
0:32:02.540,0:32:07.440
And the SZ, actually, yay, they follow,[br]right? In November, it was "Abgasskandal",
0:32:07.440,0:32:10.230
in December, it's "Volkswagen-[br]Abgasskandal".
0:32:10.230,0:32:15.830
The only problem is that even in 2000,[br]BMW was cought cheating on the Motorrad.
0:32:15.830,0:32:21.190
So this is 15 years ago. 15 years ago[br]BMW actually put the same code
0:32:21.190,0:32:30.460
which we are now seeing in Volkswagen[br]into their ECUs for the F 650 motorcycle.
0:32:30.460,0:32:35.980
And we will see again here the same 34,[br]in this case, -fold increase
0:32:35.980,0:32:39.820
in between real use and test bench use.
0:32:39.820,0:32:44.559
Now, honestly, they've been caught, they've[br]been caught earlier, and they fixed it.
0:32:44.559,0:32:48.690
So in 2001, they actually[br]brought a new version
0:32:48.690,0:32:52.720
and apparantly that didn't have[br]this cheat code anymore.
0:32:52.720,0:32:56.960
But here we see a pattern again:[br]too little time for development,
0:32:56.960,0:32:59.710
too little money willing[br]to be spent on this,
0:32:59.710,0:33:01.960
so engineers try to trick.
0:33:01.960,0:33:04.260
When you get caught,[br]and you get caught early
0:33:04.260,0:33:07.080
nobody probably of you[br]remember this here.
0:33:07.080,0:33:09.830
It's fine, it kinda fades away into history.
0:33:09.830,0:33:13.929
If you're Volkswagen, if you have[br]11 million cars out of there,
0:33:13.929,0:33:15.600
you have a big problem.
0:33:15.600,0:33:20.460
Okay, I'll skip this one, it's really[br]nice, you can see it in the slides.
0:33:20.460,0:33:25.809
But I have to go to this here[br]to give Felix enough time.
0:33:25.809,0:33:27.940
So how does component development work?
0:33:27.940,0:33:34.299
There's a huge set of legal frameworks.[br]It's a very structured top-down process.
0:33:34.299,0:33:39.370
You get requirements from the people[br]that represent the market in the company,
0:33:39.370,0:33:42.929
you get requirements from the CFO,[br]from the finance director.
0:33:42.929,0:33:47.669
And these are broken down into documents[br]which are more than a thousand pages long.
0:33:47.669,0:33:53.870
And there's every single detail that[br]could exist in this ECU written out.
0:33:53.870,0:33:58.400
There's a piece of paper[br]for everything it does.
0:33:58.400,0:34:03.220
Everything. There's not a bit in this thing[br]which is not pushed down
0:34:03.220,0:34:06.650
into a very hard set of requirements.
0:34:06.650,0:34:12.418
This is then put into a tool,[br]often Rational DOORS by IBM or something,
0:34:12.418,0:34:15.109
and then every time something changes[br]this is documented.
0:34:15.109,0:34:18.228
There's a complete paper trail, right?
0:34:18.228,0:34:20.690
So that means unless there[br]will be a cover-up,
0:34:20.690,0:34:23.829
unless we're not given all the information[br]as a public,
0:34:23.829,0:34:29.190
there's no way Volkswagen cannot find out[br]who did exactly what at what point in time,
0:34:29.190,0:34:31.579
which level of management was involved.
0:34:31.579,0:34:36.219
Because every step of the development goes[br]through a Q-Gate, a Quality Gate.
0:34:36.219,0:34:39.799
There's managers sitting there and they're[br]approving everything it does,
0:34:39.799,0:34:43.059
every progress that has been made,[br]and they're getting reports,
0:34:43.059,0:34:45.899
at least bi-weekly, on the progress.
0:34:45.899,0:34:51.159
And these reports go up the ladder, they[br]are copied to the next levels of management.
0:34:51.159,0:34:55.498
So this is a fully transparent process and[br]this is a fully top-down driven process.
0:34:55.498,0:35:00.079
It is completely impossible that you have[br]an engineer that sits there and says, like,
0:35:00.079,0:35:05.470
"Well, I wanna cheat," and does the code.[br]There's no motivation for him to do either.
0:35:05.470,0:35:11.140
He doesn't get any money for it, he would[br]only be risking his career, so he won't do.
0:35:11.140,0:35:14.789
And this is why we have paper trails,[br]and this is why engineers have written down,
0:35:14.789,0:35:17.720
"I'm doing this because my[br]manager told me to do this."
0:35:17.720,0:35:22.690
And this is why you have Bosch sending[br]a letter in 2007 to Volkswagen which says,
0:35:22.690,0:35:27.700
"We delivered you this code you[br]requested. We're your supplier, we do.
0:35:27.700,0:35:30.680
But if you send it into production[br]it will be illegal."
0:35:30.680,0:35:33.370
And they did.
0:35:35.170,0:35:40.989
So this is how actually this[br]exhaust system works.
0:35:40.989,0:35:44.109
And this is a little bit important to[br]understand what Felix is now doing
0:35:44.109,0:35:50.299
and showing you how the ECU[br]that manages this all works.
0:35:50.299,0:35:53.190
To the left would be the engine,[br]to the right is the exhaust,
0:35:53.190,0:35:56.920
the end of the exhaust[br]where the remainders come out.
0:35:56.920,0:36:02.739
And the first thing is,[br]you have diesel oxid cathalytic
0:36:02.739,0:36:06.729
and it basically takes out ...[br]The interesting stuff here is CO,
0:36:06.729,0:36:12.569
so carbon oxide, and PM, the[br]particle mass, through 98%, 50%.
0:36:12.569,0:36:17.329
The hydrocarbonides before that,[br]they just kind of don't go through
0:36:17.329,0:36:21.539
the rest of the process anymore.
0:36:21.539,0:36:27.220
Then you have a filter that basically[br]traps all of the diesel particles,
0:36:27.220,0:36:29.979
the stuff that causes[br]cancer in your lungs.
0:36:29.979,0:36:35.400
But you have to burn them out at some[br]point in time, about every 700 km,
0:36:35.400,0:36:36.989
when there have been enough collected.
0:36:36.989,0:36:38.460
So it's a bit a trick, right?
0:36:38.460,0:36:44.059
The trick is: You collect them so they[br]don't exit the exhaust
0:36:44.059,0:36:47.970
but at some point in time you have to burn[br]them again, so they do exit the exhaust.
0:36:47.970,0:36:52.729
Now the positive thing here is,[br]they get larger, and the larger they are,
0:36:52.729,0:36:58.670
the less risk they — at least as much[br]as we know — cause as a health hazard.
0:36:58.670,0:37:04.700
So this is the DPF here. And then at the end,[br]this is the really interesting thing,
0:37:04.700,0:37:07.579
this is what most of the[br]scandal now focuses on:
0:37:07.579,0:37:10.749
There's a selective catalytic reduction.
0:37:10.749,0:37:14.109
And what this thing does is,[br]it does reduce the particle mass,
0:37:14.109,0:37:16.099
it does reduce the particles.[br]That's nice.
0:37:16.099,0:37:23.239
But the interesting thing is NOx.[br]It goes against this to about 90%.
0:37:23.239,0:37:26.650
So this is what it is made for.
0:37:26.650,0:37:36.190
It basically injects urea into the airflow[br]and helps to reduce the NOx content
0:37:36.190,0:37:41.940
by creating by-products[br]which are mostly water
0:37:41.940,0:37:44.809
that comes out the end of the exhaust.
0:37:44.809,0:37:47.670
And this is the system, this is a very[br]complex technical system
0:37:47.670,0:37:51.839
that has to be managed,[br]and this is managed by an ECU.
0:37:51.839,0:37:56.440
This ECU which they selected to do this,[br]and everybody does, is the engine ECU.
0:37:56.440,0:37:59.890
Because to the left of the diagram before[br]was this big engine, you didn't see it,
0:37:59.890,0:38:04.960
it fell off the diagram, but that's[br]actually the fan blowing into the system.
0:38:04.960,0:38:10.729
So this is what you want to manage to[br]actually control what happens there.
0:38:10.729,0:38:15.989
Now this thing is quite[br]a sophisticated processor,
0:38:15.989,0:38:20.170
it's about the most complex device[br]outside multimedia and entertainment
0:38:20.170,0:38:25.400
which we find in the car, and it is a[br]very proprietory thing
0:38:25.400,0:38:28.150
because it contains a[br]physical model of engines.
0:38:28.150,0:38:32.229
So there have been hundreds, if not[br]thousands of engineers sitting there
0:38:32.229,0:38:37.309
and modelling how an engine works,[br]really physically modelling it.
0:38:37.309,0:38:41.789
And the things that an OEM —[br]an original equipment manufacturer,
0:38:41.789,0:38:44.859
a car maker — can actually tweak[br]are variables.
0:38:44.859,0:38:49.049
They can say,[br]"My engine has this and this size,
0:38:49.049,0:38:51.890
my combustion cycle looks like this and that."
0:38:51.890,0:38:55.309
But the code itself is opaque to the OEM.
0:38:55.309,0:39:02.369
It's a proprietory product which you can[br]buy from Continental, or Bosch, or so.
0:39:02.369,0:39:06.849
And there's about 20,000 variables[br]which you can tune.
0:39:06.849,0:39:12.920
And this thing is simulated and tested[br]to death. Because it is hugely important.
0:39:12.920,0:39:17.349
Because you have this machine here[br]that has like 100, 200 horsepowers
0:39:17.349,0:39:21.509
and if you steer it wrong it will blow up,[br]and it will blow up really hard.
0:39:21.509,0:39:29.039
So this is why this thing is about the best[br]tested piece of software you will ever find.
0:39:29.039,0:39:34.729
Which also again means there's everything[br]documented, everything is written down,
0:39:34.729,0:39:39.029
everything is seen by everybody[br]who's working with these,
0:39:39.029,0:39:41.979
whether it's in development,[br]whether it's in integration,
0:39:41.979,0:39:45.720
whether it's in the plants that[br]flash these things, and so on.
0:39:45.720,0:39:47.630
There's nothing secret here in this, right?
0:39:47.630,0:39:51.319
The functions which are there are[br]actually there to be seen,
0:39:51.319,0:39:58.560
well, seen if they are named apparantly, and[br]that is something that Felix will talk about.
0:40:00.754,0:40:09.744
audience applauds
0:40:09.744,0:40:13.190
F: Thank you. Hey, okay.[br]So I will do the second part of this talk.
0:40:13.190,0:40:15.029
I'm Felix, by the way.
0:40:15.029,0:40:18.400
So my motivation with this[br]was a little bit different.
0:40:18.400,0:40:25.749
I'm curious, and, I mean, we can find a lot[br]of source material for this whole scandal.
0:40:25.749,0:40:28.249
We can find a lot of[br]information in the press,
0:40:28.249,0:40:31.759
a lot of information in the[br]Volkswagen press releases.
0:40:31.759,0:40:36.779
However, it should be easier[br]because all the cars are there,
0:40:36.779,0:40:42.950
the 11 million cars are out there[br]that have the cheat code in them.
0:40:42.950,0:40:47.420
And we are hackers, and we know code,[br]and the truth is in the code.
0:40:47.420,0:40:52.660
So my approach was, well,[br]let's take a car, let's take it apart,
0:40:52.660,0:40:55.729
let's take the firmware out of it,[br]let's throw it in a disassembler,
0:40:55.729,0:40:59.069
maybe get some measurements, and[br]then look at what the car is actually doing
0:40:59.069,0:41:04.799
instead of relying on all of this[br]second-hand, third-hand information.
0:41:04.799,0:41:06.660
So what do we need for this approach?
0:41:06.660,0:41:10.969
So first of all, we need a car[br]that's affected.
0:41:10.969,0:41:14.910
You need to drive that car somehow,[br]and driving a car on an open road
0:41:14.910,0:41:18.160
can be dangerous if you have to follow[br]particular driving cycles.
0:41:18.160,0:41:21.789
So there's a "dyno" you can put the car on[br]and then you can just drive
0:41:21.789,0:41:23.599
without the car physically moving.
0:41:23.599,0:41:26.309
The wheels are moving,[br]but the car isn't moving.
0:41:26.309,0:41:28.069
And this is what other people have done,
0:41:28.069,0:41:31.249
and they have taken very interesting[br]measurements out of this.
0:41:31.249,0:41:34.170
However, we as hackers,[br]we can go one step further.
0:41:34.170,0:41:37.869
We can take a look at the ECU itself.
0:41:37.869,0:41:45.469
And not only that, we can also ask[br]other people who worked with these things
0:41:45.469,0:41:50.029
and may be able to get[br]more information about them.
0:41:50.029,0:41:51.799
I will talk about this in a minute.
0:41:51.799,0:41:58.039
So first of all, this is my car, luckily that[br]car was affected by the recall.
0:41:58.039,0:42:02.039
So I was very happy when I got the letter[br]telling me I have to go to the shop in January
0:42:02.039,0:42:05.559
and get a firmware update because[br]firmware updates are exciting, right?
0:42:05.559,0:42:09.969
I love updating things,[br]so updating a car seems great.
0:42:09.969,0:42:14.279
Yeah, it sucked that my car was putting out[br]more emissions than it should have,
0:42:14.279,0:42:17.599
but otherwise, it gave me the chance[br]to actually look at the car.
0:42:17.599,0:42:24.599
I mean, I could have rented a car or[br]something, but that makes it much easier.
0:42:24.599,0:42:28.460
I also went on a dyno with my car.[br]On a dyno, there are no speed limits
0:42:28.460,0:42:32.509
or no people to run over when you just[br]have to keep a constant speed or something,
0:42:32.509,0:42:35.359
so it makes things much easier.
0:42:35.359,0:42:39.969
And I talked about ripping apart[br]my car and disassembling it.
0:42:39.969,0:42:44.410
I didn't really want to do that,[br]so what I did instead was what I always do:
0:42:44.410,0:42:49.789
I go to eBay and I bought an extra ECU.
0:42:49.789,0:42:54.929
Here it is, maybe you can show it?
0:42:54.929,0:42:58.229
You can go here after the talk[br]and take a look at it.
0:42:58.229,0:43:03.989
This is the ECU. This here is the main CPU[br]that also includes the flash.
0:43:03.989,0:43:09.509
On the other side there are the power drivers[br]that drive the actual stuff in the car.
0:43:09.509,0:43:12.460
And then there's other[br]watchdog circuits and so on.
0:43:12.460,0:43:14.760
Okay, thank you.
0:43:16.970,0:43:23.180
So, the ECU was built by Bosch,[br]it's an EDC17C46,
0:43:23.180,0:43:25.299
that's the name of the hardware.
0:43:25.299,0:43:29.099
And it can easily be obtained on eBay,[br]and you can put it on your desk,
0:43:29.099,0:43:31.599
you apply 12 volt to it and then it boots.
0:43:31.599,0:43:34.279
It will complain about a lot of[br]sensors being missing and so on
0:43:34.279,0:43:37.650
but you can see it executing code.
0:43:37.650,0:43:43.150
And it doesn't have the very same[br]firmware as my car, but it's very close.
0:43:43.150,0:43:46.479
The flash chip is unfortunately in the[br]same pakage as the main CPU,
0:43:46.479,0:43:49.519
which is an Infineon TriCore chip,
0:43:49.519,0:43:51.779
which is apparantly only used[br]in automotive equipment,
0:43:51.779,0:43:55.690
or at least I'm only aware of it[br]being used there.
0:43:55.690,0:43:59.380
And I was able to dump the flash by[br]attacking the hardware
0:43:59.380,0:44:03.989
and exploiting a bug in the hardware[br]that I haven't found documented anywhere,
0:44:03.989,0:44:06.359
but it was not that complicated.
0:44:06.359,0:44:09.999
And then I had a firmware dump, I had a[br]2 megabit binary,
0:44:09.999,0:44:12.329
and I throw it in a disassembler.
0:44:12.329,0:44:16.910
And what we see is interesting because[br]the code is written very different
0:44:16.910,0:44:18.829
from other code that we know.
0:44:18.829,0:44:20.930
So usually, code has[br]a lot of flow control
0:44:20.930,0:44:24.400
and usually more or less[br]resembles spaghetti code.
0:44:24.400,0:44:26.769
This was the exact opposite.
0:44:26.769,0:44:31.739
It's more like someone took electrical[br]schematics and put them into code.
0:44:31.739,0:44:35.529
There's a set of input signals,[br]there's a set of processing on it,
0:44:35.529,0:44:37.009
and there's a set of output signals.
0:44:37.009,0:44:42.180
That gets updated every 10 ms or[br]once per rotation depending on processoids.
0:44:42.180,0:44:46.759
Really interesting way of writing software[br]and building this.
0:44:46.759,0:44:52.279
Also it's very data-driven, so a large[br]part of the firmware is not code but is data.
0:44:52.279,0:44:55.829
All of the computations,[br]they don't use constants at all,
0:44:55.829,0:44:58.849
they always refer to something[br]from the data section.
0:44:58.849,0:45:07.109
As Daniel said, Bosch writes this code,[br]the code is not directly visible to Volkswagen,
0:45:07.109,0:45:10.150
but they have visibility into this data,[br]and they know what the data does.
0:45:10.150,0:45:12.210
They have tools to change the data.
0:45:12.210,0:45:16.839
Volkswagen and other companies[br]can customize this,
0:45:16.839,0:45:20.150
really they cannot just customize it,
0:45:20.150,0:45:24.089
they can change the whole[br]behaviour of this ECU
0:45:24.089,0:45:29.559
by changing just the data, not the code.
0:45:29.559,0:45:34.369
The ECU really is a small embedded machine[br]in your car that takes care of the engine,
0:45:34.369,0:45:38.849
it's an Engine Electronic Control Unit,[br]there are multiple names for it.
0:45:38.849,0:45:42.229
The most important thing that it does is[br]that it takes sensor input,
0:45:42.229,0:45:46.180
for example the throttle, and then it[br]applies control to the system.
0:45:46.180,0:45:49.420
For example it calculates the amount of fuel[br]to inject, the amount of air to inject
0:45:49.420,0:45:54.539
to make the motor running at the speed[br]you want it to run.
0:45:54.539,0:45:56.630
These days it's much more complicated.
0:45:56.630,0:46:02.400
One important thing the ECU does[br]these days is emission control.
0:46:02.400,0:46:06.710
This is why we would expect to find the[br]"cheat code", the code that cheats
0:46:06.710,0:46:09.710
that Volkswagen used to[br]cheat in the whole thing,
0:46:09.710,0:46:13.430
we would expect to find it in the ECU.
0:46:13.430,0:46:16.880
Now taking a look at[br]two megabyte firmware binaries
0:46:16.880,0:46:19.769
that doesn't have any visible strings in it,
0:46:19.769,0:46:22.880
it's kind of painful if you're just suscepting[br]a code analysis.
0:46:22.880,0:46:30.339
So what I did was to do realtime logging.
0:46:30.339,0:46:35.279
You can actually read data from your ECU[br]by plugging into this OBD-II port
0:46:35.279,0:46:36.999
which is next to your steering wheel.
0:46:36.999,0:46:40.180
And while the engine is running you can[br]read out certain data.
0:46:40.180,0:46:43.400
Usually you can read out boring data[br]like RPM, and speed,
0:46:43.400,0:46:46.640
and some things that the[br]vendor wants you to see.
0:46:46.640,0:46:48.910
But there's also a mode that's[br]a little bit hidden,
0:46:48.910,0:46:51.089
but you can get pretty easily into it,
0:46:51.089,0:46:54.559
where you can read by address,[br]where you can just read the whole memory.
0:46:54.559,0:46:58.630
Well, not everything.[br]Some security data is locked out.
0:46:58.630,0:47:02.670
But the data we are interested in,[br]we can read that memory.
0:47:02.670,0:47:07.569
Now we still need to understand[br]where the interesting stuff is.
0:47:07.569,0:47:09.630
We can disassemble the firmware,[br]and that's all fine.
0:47:09.630,0:47:13.120
We can also get a little help[br]from something called "A2L files".
0:47:13.120,0:47:17.799
The chip tuners use them extensively[br]when they change the mappings,
0:47:17.799,0:47:20.640
they want to optimize an engine[br]for a different goal,
0:47:20.640,0:47:24.519
for example for more power instead of[br]long lifetime, or something.
0:47:24.519,0:47:29.219
They change things in the ECU firmware.
0:47:29.219,0:47:34.029
They do reverse engineer a lot,[br]but they also got these files.
0:47:34.029,0:47:37.109
And I'm not sure how they got them,[br]but they are out there.
0:47:37.109,0:47:40.950
And if you use the right Google terms[br]you will find them.
0:47:40.950,0:47:42.920
They are specific to each firmware.
0:47:42.920,0:47:45.029
I wasn't able to find one for[br]my actual firmware
0:47:45.029,0:47:49.829
but I was able to find one for[br]firmware that is close to mine.
0:47:49.829,0:47:53.369
And if you look into this file,[br]what you see is the symbol names,
0:47:53.369,0:47:54.749
it's basically a fancy map file.
0:47:54.749,0:48:00.940
You see the symbol names, you see a[br]mostly German description of that symbol,
0:48:00.940,0:48:06.489
you see a real-use unit, and you see the[br]adress in memory that we can read at.
0:48:06.489,0:48:12.489
So with the help of these files we can read[br]out almost any internal state in the ECU.
0:48:12.489,0:48:16.650
We still have to make sense out of that,[br]but at least we know where the data is
0:48:16.650,0:48:20.420
and what to look for.
0:48:20.420,0:48:26.119
It's surprising how complex an ECU is.[br]For example, this thing, what does it display?
0:48:26.119,0:48:32.209
Everybody would say it's a function of RPM,[br]it shows you how fast the engine is running.
0:48:32.209,0:48:37.249
Well, it's not quite the case,[br]and if we look careful we see that
0:48:37.249,0:48:42.180
this code is post-processing[br]the RPM signal.
0:48:42.180,0:48:47.019
It's 12 kilobyte of densely written code[br]that has a lot of internal state
0:48:47.019,0:48:50.250
that tries to make the RPM value,
0:48:50.250,0:48:52.739
convert it to something[br]that the customer wants to see.
0:48:52.739,0:48:57.180
For example, you want your idle speed[br]to be stuck at 780, you don't want it to oscillate.
0:48:57.180,0:49:01.039
But in reality it does,[br]and this code takes away all of that
0:49:01.039,0:49:05.890
and makes it flat 780.
0:49:05.890,0:49:10.049
You realize probably at this point that there[br]is a lot of cheating that could go on here
0:49:10.049,0:49:12.459
without most people noticing.
0:49:12.459,0:49:17.839
You don't really believe that the speedometer[br]in your car displays your actual speed, right?
0:49:17.839,0:49:22.190
It displays something related to speed ...
0:49:22.970,0:49:25.209
But let's get back to topic.
0:49:25.209,0:49:29.349
Selective Catalytic Reduction is the process[br]of, well, if you don't have it
0:49:29.349,0:49:34.400
you get a lot of NOx, of nitrogen oxides[br]at the end of the exhaust.
0:49:34.400,0:49:36.940
That's bad, you don't want that.
0:49:36.940,0:49:42.109
There is one way of getting rid of this,[br]is to add an SCR catalyst.
0:49:42.109,0:49:46.099
And the SCR catalyst —[br]I simplified this a lot,
0:49:46.099,0:49:48.929
you can find a lot more information[br]about this —
0:49:48.929,0:49:56.839
SCR is a process that reduces the NOx[br]using something called DEF,
0:49:56.839,0:50:00.599
or AdBlue is a term for it.[br]It's some fluid that you put in there.
0:50:04.029,0:50:06.859
Basically it's an Urea/water solution.
0:50:06.859,0:50:13.019
And the AdBlue, at a high temperature,[br]converts to Ammonia
0:50:13.019,0:50:16.469
and then it reacts with the NOx[br]to nitrogen and water.
0:50:16.469,0:50:21.809
Which is great because that's not[br]in any way harmful to us.
0:50:21.809,0:50:26.440
However, there's a problem here because[br]the dosage of the AdBlue needs to be correct
0:50:26.440,0:50:28.630
and it's very hard to do.
0:50:28.630,0:50:33.690
If we dose too little of that[br]the conversion is not perfect
0:50:33.690,0:50:36.069
and we will still get[br]a lot of NOx at the output.
0:50:36.069,0:50:38.410
Which is better than not doing anything.
0:50:38.410,0:50:41.619
It's not perfect,[br]but it's not more harmful than before.
0:50:41.619,0:50:45.359
However, if you put in[br]too much of the AdBlue
0:50:45.359,0:50:50.089
what you get at the output is ammonia,[br]and you really don't want that.
0:50:50.089,0:50:54.769
So the primary goal of emission control[br]is, if you have the SCR system,
0:50:54.769,0:50:58.630
is to eliminate as much[br]as possible of the NOx
0:50:58.630,0:51:03.170
and minimize the amount of ammonia[br]that comes out of the exhaust pipe.
0:51:03.170,0:51:06.640
Ammonia is NH3.
0:51:06.640,0:51:10.589
Calculating the right dosage works[br]with a model again.
0:51:10.589,0:51:13.459
They modeled everything that happens[br]in the exhaust process,
0:51:13.459,0:51:17.390
they have a model of the catalyst,[br]they have a model of the internal state,
0:51:17.390,0:51:23.339
they do have a number of sensors and[br]outputs from the other models
0:51:23.339,0:51:25.180
that tell them a lot of values.
0:51:25.180,0:51:30.349
And the model uses this with a lot of[br]internal storage, internal state.
0:51:30.349,0:51:34.619
And the model then calculates[br]the amount of AdBlue to dose
0:51:34.619,0:51:42.839
to convert as much NOx as possible[br]without leaking any ammonia.
0:51:42.839,0:51:47.180
The way things usually work in an ECU is,[br]there's one system that controls things
0:51:47.180,0:51:50.079
and there's another system[br]that monitors things.
0:51:50.079,0:51:54.200
It's independent from the main system,[br]it tries to be as independent as possible.
0:51:54.200,0:51:57.849
It's still running on the same hardware[br]but it's not sharing a lot of code.
0:51:57.849,0:52:04.390
There is an efficiency monitoring scheme that,[br]if the conversion is not good enough anymore,
0:52:04.390,0:52:07.930
it will flag this as an OBD-II error
0:52:07.930,0:52:10.209
and you will see your[br]"check engine" light going on,
0:52:10.209,0:52:13.999
and then you go to the shop, and the shop[br]will diagnose your car and will fix this,
0:52:13.999,0:52:18.549
for example if your catalyst is broken.
0:52:18.549,0:52:21.749
Based on the test results we would have[br]expected this efficiency monitoring
0:52:21.749,0:52:27.219
to actually flag the inefficiencies.[br]But it didn't.
0:52:27.219,0:52:30.410
It turns out the main model[br]doesn't always work.
0:52:30.410,0:52:33.719
There are some operating conditions[br]where the main model is not sufficient,
0:52:33.719,0:52:38.499
it has certain bounds where it works,[br]and outside of these conditions —
0:52:38.499,0:52:44.630
for example if the engine is too hot or if[br]the exhaust mass is too large —
0:52:44.630,0:52:46.739
the model doesn't produce[br]meaningful results.
0:52:46.739,0:52:51.789
It may overdose the AdBlue,[br]and we don't want that.
0:52:51.789,0:52:55.450
There's an alternative model[br]which is much, much simpler,
0:52:55.450,0:52:58.209
and takes only a few sensory inputs,
0:52:58.209,0:53:01.579
and doesn't rely on as many variables[br]to be perfect.
0:53:01.579,0:53:04.799
It will still calculate an AdBlue dosage.
0:53:04.799,0:53:10.009
However, the main goal of this alternative[br]model is to make the exhaust processing work
0:53:10.009,0:53:17.329
in all situations without ever[br]overdosing the NH3.
0:53:17.329,0:53:22.549
They're calculating both of these models and[br]then they are selecting one of the models.
0:53:22.549,0:53:26.140
The output of the selection then controls[br]the AdBlue dosage,
0:53:26.140,0:53:29.269
the pump that injects the AdBlue[br]into the exhaust.
0:53:29.269,0:53:34.099
There's code that controls[br]which of the models to use.
0:53:34.099,0:53:39.209
There's also a statistics model that counts[br]how often each mode is selected.
0:53:39.209,0:53:43.319
Again, all of this model selection[br]depends on the data.
0:53:43.319,0:53:45.739
It's code that does the selection[br]but it depends on a lot of data,
0:53:45.739,0:53:48.910
there are parameters tought of this.
0:53:48.910,0:53:51.769
Let's take a look at the selection criteria[br]for this alternative model.
0:53:51.769,0:53:55.200
We see that a lot of these parameters[br]are dummy variables,
0:53:55.200,0:53:56.989
things that can never happen.
0:53:56.989,0:54:01.979
For example, the athmospheric pressure[br]can't be negative, that can never happen.
0:54:01.979,0:54:06.420
Or the air temperature ...[br]I hope it's never larger than that,
0:54:06.420,0:54:09.079
or smaller than 0.1K, right?
0:54:09.079,0:54:11.859
However, one thing stuck out,
0:54:11.859,0:54:17.749
and that was a check if the engine condition[br]is larger than negative temperature.
0:54:17.749,0:54:19.839
Which does not exist,[br]the temperature is always positive.
0:54:19.839,0:54:23.420
That last one is always true,
0:54:23.420,0:54:27.549
so the model that would be selected would[br]always be the alternative model.
0:54:27.549,0:54:30.150
That sounded weird and[br]I was looking at the firmware.
0:54:30.150,0:54:34.640
Maybe I understood it incorrectly,[br]or maybe I looked at the wrong place
0:54:34.640,0:54:37.630
when looking at these parameters?
0:54:37.630,0:54:42.369
But if we look at the intermediate results[br]there is a bit at a certain location
0:54:42.369,0:54:47.579
that tells us which model was selected,[br]and that bit is indeed always set.
0:54:47.579,0:54:51.249
That is weird, it sounds fishy.
0:54:51.249,0:54:57.489
Let's take a look at the statistics,[br]the car counts what model you're in.
0:54:57.489,0:55:00.309
20% of the cases[br]my car does not do dosing at all.
0:55:00.309,0:55:02.630
So I drove some time and then[br]looked at the values.
0:55:02.630,0:55:06.369
And the 20% where it doesn't do anything[br]is mostly the warm-up cycle.
0:55:06.369,0:55:09.469
But everytime it does something,[br]it's actually the alternative model
0:55:09.469,0:55:14.719
which we know does underdose NH3[br]because it doesn't want to leak ammonia.
0:55:14.719,0:55:19.449
And that makes sense because my car[br]uses much less than expected of the AdBlue.
0:55:19.449,0:55:25.529
The expected value is roughly 2.5 liters[br]per 1000 kilometers, of the AdBlue.
0:55:25.529,0:55:28.779
In my case it only used 0.6 liters[br]per 1000 kilometers.
0:55:28.779,0:55:32.469
Which is great for me because I don't have[br]to refill this tank very often.
0:55:32.469,0:55:37.319
In fact, I never had to do it,[br]the shop always does it when I'm there.
0:55:37.319,0:55:42.380
But this is fishy,[br]and let's take a look at this.
0:55:42.380,0:55:46.029
What we also see is that sometimes[br]the regular model is active,
0:55:46.029,0:55:47.859
so there must be something more.
0:55:47.859,0:55:52.539
If we look at the selection criteria we find[br]that there's an additional term there
0:55:52.539,0:55:54.959
that I haven't found before.
0:55:54.959,0:55:57.049
There's an additional condition[br]that has to be true
0:55:57.049,0:56:01.680
in order to go to the alternative model[br]that underdoses.
0:56:01.680,0:56:04.989
We look at the particular conditions[br]and we find a lot of stuff
0:56:04.989,0:56:07.739
that is related to diagnostics,[br]things they can do in the shop.
0:56:07.739,0:56:09.969
So that's definitely not[br]happening on the street.
0:56:09.969,0:56:12.959
But one of the criteria,[br]that really was weird
0:56:12.959,0:56:18.769
because it looks if the engine and fuel[br]temperature is larger than 50°C,
0:56:18.769,0:56:24.249
it looks at the athmospheric pressure[br]and if it's lower than 750m,
0:56:24.249,0:56:25.809
that must be satisfied.
0:56:25.809,0:56:29.859
If all of these conditions are satisfied[br]it will move back to the main model
0:56:29.859,0:56:33.759
that does the proper exhaust processing.[br]And one thing was really weird.
0:56:33.759,0:56:36.380
There were seven curves,[br]not all of them used,
0:56:36.380,0:56:38.789
that define an upper and a lower bound
0:56:38.789,0:56:42.799
on the distance driven[br]after a certain amount of time.
0:56:42.799,0:56:46.459
This is how it looks in disassembly.[br]I'm not sure if you can read this.
0:56:46.459,0:56:52.699
But the comments are from this A2L file[br]and they call it "acoustic function".
0:56:52.699,0:56:55.440
I'm not sure if this has anything[br]to do with acoustics.
0:56:55.440,0:57:00.900
I tried to find all the usages, and there[br]was nothing related to sound or anything.
0:57:00.900,0:57:04.180
I think it's just a name for it.
0:57:04.180,0:57:09.579
Now if we go and take a look at these[br]upper and lower bounds, we see this:
0:57:09.579,0:57:16.999
These are three curves that are defined,[br]each of them has an upper and a lower bound.
0:57:16.999,0:57:19.259
It's basically the distance
0:57:19.259,0:57:22.670
that you need to have driven[br]after a certain amount of time.
0:57:22.670,0:57:27.479
And if you ever fall out of one of these curves[br]we're switching back to the alternative model
0:57:27.479,0:57:30.859
that underdoses NH3[br]and causes the inefficiencies.
0:57:30.859,0:57:34.109
This is weird,[br]and I didn't really know what this is.
0:57:34.109,0:57:38.709
Let's get back to something[br]completely different, which is the NEDC.
0:57:38.709,0:57:43.219
We've seen this slide before,[br]the NEDC mandates you how to drive.
0:57:43.219,0:57:46.479
One thing is also interesting:[br]It mandates you that ...
0:57:46.479,0:57:50.269
You want this test at "cold-start",[br]and what's better for a cold start
0:57:50.269,0:57:54.660
than heating the car to 20°C[br]and keep it that warm until you start.
0:57:54.660,0:58:02.819
That's the "cold-start", that's the[br]cold start as defined in the law: 20°C.
0:58:02.819,0:58:10.339
This is speed over time, so to get[br]distance over time we need to integrate this.
0:58:10.339,0:58:12.869
And we get this graph.
0:58:12.869,0:58:17.479
And if we overlay what we found in the[br]firmware we get this.
0:58:17.479,0:58:27.140
audience laughs and applauds
0:58:27.140,0:58:31.660
What we can see here is that if you drive[br]the driving cycle correctly
0:58:31.660,0:58:35.660
you will exactly be in the bounds[br]of one of these curves.
0:58:35.660,0:58:37.400
And you can do this on the street,[br]you can do this everywhere.
0:58:37.400,0:58:42.779
As long as you satisfy the distance over[br]time and your car is warm enough
0:58:42.779,0:58:46.609
it will detect this in some way.
0:58:46.609,0:58:49.650
Well, you can drive this on a street,[br]but it's really dangerous
0:58:49.650,0:58:52.690
because you have to follow[br]a given speed pattern.
0:58:52.690,0:58:55.799
So i did this on a dyno,[br]I put my laptop in there,
0:58:55.799,0:58:58.849
I logged the data in real-time[br]and then displayed it.
0:58:58.849,0:59:02.959
Basically, this is what it looks like.[br]In the middle you see a bar.
0:59:02.959,0:59:06.469
You have to drive and keep this[br]middle bar in the middle,
0:59:06.469,0:59:11.410
which means you are well within this upper[br]and lower bound, and not try to escape it.
0:59:11.410,0:59:17.099
And as long as you do, one of the[br]other green boxes will tell you
0:59:17.099,0:59:22.249
that the car is still detecting this[br]as being in this cycle.
0:59:22.249,0:59:29.539
Then what I did in the end, I stayed in the[br]cycle for a while and I logged all the data.
0:59:29.539,0:59:32.089
At the end I would just hit[br]a constant speed
0:59:32.089,0:59:36.130
which would eventually get me[br]out of the conditions.
0:59:36.130,0:59:39.509
This is the log that I made.
0:59:39.509,0:59:41.880
On the first graph you see[br]the vehicle speed,
0:59:41.880,0:59:45.049
you see how I tried to follow the NEDC[br]more or less successfully.
0:59:45.049,0:59:49.559
On the second graph you see[br]the distance over time,
0:59:49.559,0:59:54.680
you see that I stay within the bounds[br]enforced by the firmware.
0:59:54.680,0:59:56.739
You an also see on the third graph—
0:59:56.739,0:59:59.179
this is the actual signal at the AdBlue pump—
0:59:59.179,1:00:02.130
that it actually doses[br]quite a lot of AdBlue.
1:00:02.130,1:00:05.630
It calculates the amount of AdBlue to dose[br]based on the model output
1:00:05.630,1:00:09.420
which you see in graph 5 and 6.
1:00:09.420,1:00:14.839
By the way, graph 4 is the actual NOx[br]emitted by the engine based on their model.
1:00:14.839,1:00:20.789
That's the RML, their mission model then[br]calculates the amout of the dosing to happen.
1:00:20.789,1:00:25.880
As we see, as long as we stay within the[br]limits enforced that match the NEDC
1:00:25.880,1:00:29.579
everthing is good[br]and a lot of AdBlue is dosed.
1:00:29.579,1:00:32.749
And then, in the end, I drove too fast.
1:00:32.749,1:00:35.979
And you can see in the second graph[br]that I crossed the upper bar,
1:00:35.979,1:00:38.249
the blue line goes[br]over the red line, right?
1:00:38.249,1:00:40.989
You can see that the car[br]immediately detects this,
1:00:40.989,1:00:44.529
that I'm no longer in the driving cycle.
1:00:44.529,1:00:52.259
The interesting part you see here is the[br]effect on the AdBlue dosing, which is here.
1:00:52.259,1:00:57.619
It immediately stops doing the dosing.[br]And you can see in the model below
1:00:57.619,1:01:01.670
the model still calculates that[br]AdBlue should be dosed.
1:01:01.670,1:01:04.630
But after they have the max,[br]after they switch the model
1:01:04.630,1:01:08.709
and switch to the alternative model,[br]the alternative model just outputs zeroes,
1:01:08.709,1:01:11.869
it doesn't dose anything.
1:01:11.869,1:01:14.359
This shows that when we're[br]following the cycle
1:01:14.359,1:01:17.539
everything is fine,[br]enough Urea is dosed,
1:01:17.539,1:01:24.599
and then once we leave the cycle,[br]there's a severe reduction in the dosing.
1:01:24.599,1:01:26.949
And it's all based on[br]detecting this driving cycle.
1:01:26.949,1:01:29.180
Two more slides.[br]A: Two more slides.
1:01:29.180,1:01:32.349
F: Two more slides.[br]A: Two more slides, here we go!
1:01:32.349,1:01:38.299
audience laughs and applauds
1:01:38.299,1:01:40.979
I have to be clear[br]on the limitations here.
1:01:40.979,1:01:43.599
All of this was looking at[br]disassembled code and so on,
1:01:43.599,1:01:47.559
I could have done something wrong here,[br]so take this with a grain of salt.
1:01:47.559,1:01:50.979
We couldn't do NOx measurements[br]on the dyno, unfortunately.
1:01:50.979,1:01:55.569
And I have to stress: We looked at one[br]particular car that uses SCR processing,
1:01:55.569,1:01:57.660
not all of the affected cars are doing this,
1:01:57.660,1:01:59.719
there are some other[br]mechanisms in the other cars.
1:01:59.719,1:02:02.559
And I looked at a car[br]for the German market,
1:02:02.559,1:02:05.569
at least the curves have to be different[br]for the other markets.
1:02:05.569,1:02:10.900
Let's reenumerate the results—[br]and this is my last slide.
1:02:10.900,1:02:16.749
Most of the time, on a regular car,[br]a nonstandard treatment mode is active
1:02:16.749,1:02:20.799
that is not as efficient[br]as the real mode that is implemented.
1:02:20.799,1:02:23.309
We can show the code[br]that is responsible for this:
1:02:23.309,1:02:26.269
This is this negative temperature limit[br]that they look at
1:02:26.269,1:02:29.809
which doesn't make any sense and[br]always selects the alternative mode.
1:02:29.809,1:02:32.630
And we can see, in the logs,[br]the state selection bit,
1:02:32.630,1:02:38.359
we can see the counters that count[br]that the alternative model is active.
1:02:38.359,1:02:42.019
We can see that there's an AdBlue[br]underdosing in this state
1:02:42.019,1:02:44.920
which causes the inefficient[br]NOx conversions,
1:02:44.920,1:02:49.609
that's what we've seen before when[br]people put the car on the dyno.
1:02:49.609,1:02:52.939
We know that the efficiency checks[br]are only enabled in the main mode
1:02:52.939,1:02:57.019
and the car does exceed the limits.
1:02:57.019,1:03:03.400
This shows how the alternate model is[br]selected where it doses too little AdBlue
1:03:03.400,1:03:06.569
and causes the inefficient conversion.
1:03:06.569,1:03:10.209
We can see that if we[br]follow the driving cylce,
1:03:10.209,1:03:12.459
the minimum temperature and[br]the distance over time,
1:03:12.459,1:03:14.869
we will see that it switches[br]to the main model
1:03:14.869,1:03:17.640
that should have been active[br]all the time.
1:03:17.640,1:03:20.289
We can show the code[br]that's responsible for that,
1:03:20.289,1:03:23.529
the driving cycle detection that uses[br]the upper bound and the lower bound.
1:03:23.529,1:03:28.630
We can extract the exact limits, overlay[br]the NEDC data and see that there's a match.
1:03:28.630,1:03:33.680
We can, if we do this actually on a dyno,[br]we can see how it switches the SCR state.
1:03:33.680,1:03:37.319
We can show the effect on the DEF dosing,[br]on the AdBlue dosing.
1:03:37.319,1:03:41.499
As you've seen on the slide before,[br]as soon as we switch out of the driving cycle
1:03:41.499,1:03:48.170
into the street mode,[br]the dosing will get close to zero.
1:03:48.170,1:03:50.910
Once you're back in the main model[br]all the efficiency checks are enabled,
1:03:50.910,1:03:54.519
for example to take better Urea.
1:03:54.519,1:03:56.979
So the efficiency checks are there,[br]but they are not active
1:03:56.979,1:04:00.680
because the car is forced to run[br]in the alternative model.
1:04:00.680,1:04:04.709
These results are all in line[br]with the Volkswagen press releases.
1:04:04.709,1:04:07.859
These are basically just the details[br]as extracted from the firmware
1:04:07.859,1:04:10.450
to show you the background.
1:04:10.450,1:04:12.330
Thank you.
1:04:12.330,1:04:16.190
audience applauds
1:04:16.190,1:04:20.229
A: Wow![br]Thank you very much, Daniel and Felix.
1:04:20.229,1:04:35.919
audience applauds
1:04:37.609,1:04:40.269
I'm really sorry,[br]but we have to clear the stage.
1:04:40.269,1:04:42.609
There is not going to be time[br]for the Q&A session.
1:04:42.609,1:04:46.650
Do that down there. I'm sure that a few[br]people just come down,
1:04:46.650,1:04:49.670
grab you and ask questions.[br]Unfortunately, we can't do that.
1:04:49.670,1:04:54.059
I have to close it in exactly four seconds[br]over here because we have to go off the stream.
1:04:54.059,1:04:57.529
Thank you very much Felix,[br]thank you very much Daniel.
1:04:57.529,1:05:03.387
[br]F: Thank you.
1:05:03.387,1:05:05.676
♪ postroll music ♪
1:05:05.676,1:05:11.000
subtitles created by c3subtitles.de[br]Join, and help us!