bill shoe wrote:I define “defeat device” as something that changes the real-world behavior of a system compared to its behavior during regulatory testing, and this change must occur within the range of evaluation that occurred during regulatory certification.
If you drive the VW diesel on-road with the same driving cycle (range of evaluation) that was used on the EPA chassis dyno then you get different emissions results (behavior of the system). It’s a defeat device, VW is guilty, simple.
F1 case 1, basic car
FIA test rig: An upward force is applied to the front center of the t-tray. The force is increased until it reaches the maximum test force, and the resulting deflection is within the allowable limit.
On-track: When the upward force on the front center of the t-tray is less than the maximum force from the test rig, the deflection is the same as on the test rig. When the force is larger than the test rig’s maximum force, the deformation continues roughly linearly.
The car’s on-track behavior does not deviate from its regulatory behavior. This car doesn’t have a defeat device.
F1 case 2, non-linear car
FIA test rig: Same as case 1.
On-track: When the upward force on the front center of the t-tray is less than the maximum force from the test rig, the deflection is the same as on the test rig. When the force is larger than the test rig’s maximum force, the deformation continues at an increasing rate (perhaps via a mechanism that buckles when a certain force is exceeded).
The car’s on-track behavior sometimes deviates from its regulatory behavior, but only when outside the range of evaluation (range of force) that occurred on the FIA test rig. This characteristic is not a defeat device.
F1 case 3, Hinged t-tray/floor system that’s constrained when car lifted by FIA test rig
FIA test rig: Same as case 1.
On-track: When the upward force on the front center of the t-tray is less than the maximum force from the test rig, the deflection is much greater than what occurred on the test rig.
This car deviates from its regulatory behavior by large amounts, and it deviates within the range of evaluation (range of force) that occurred on the test rig. By definition this characteristic is a defeat device, and is analogous to the VW defeat device.
If you don’t like this explanation then define “defeat device” in a way that can be applied to both road car emissions and F1 car technical regulations.
Pls allow me to reference my own post, awkward I know.
There is growing talk of emissions levels during regulatory dyno-test vs real-world driving (and not just VW). Also growing use of terms like defeat-device.
The U.S EPA test cycle has been described as your grandmother driving your car. The European test cycle has been described as your grandmother on sedatives driving your car. The accel rates are slow, and the transients are quite gentle. If you take a car, essentially any modern emission-controlled car, and drive it in the real word then it will emit more pollution (per mile or whatever) than the dyno test. This is because engines and transmissions are tuned to provide efficiency and low-pollution during very light use, but strong performance (fuel consumption be damned) during harder use. The average consumer drives significantly harder than the EPA or Euro test cycle.
Is there anything in U.S./EPA legal code that says in a clear and simple way: "The intent of the dyno test is to represent real-world conditions, therefore if a car makes more real-world pollution than dyno-test pollution then this situation is de-facto a violation of the law." If there was a legal obligation to perform the same in real-world vs dyno-test then I would think there should be some statement to that effect. An absence of such an obvious statement suggests there is no obligation to be identical in dyno-testing vs real-world. The paragraph that describes defeat-devices is not a simple, clear statement that a car must always produce identical dyno-test vs real-world results regardless of differences in how it's driven in the two situations.
If deviation between dyno-test vs real-world was a de-facto violation of law, then would a carmaker violate the law by selling a car to an average consumer with an average (harder than dyno test) driving style? Logically the answer would have to be yes, therefore I can't see any plausible interpretation that differences in dyno vs reality necessarily violates law. They could, but not necessarily.
Going back to my F1 examples, my best understanding of the VW situation is that they are analogous to my F1 case 3, because they have a defeat-device. In contrast, the other 97% of cars (I'm hoping nobody else is doing overt defeat-devices like VW) are analogous to my F1 case 2 where real-world results are different from regulation dyno-testing simply because real-world use exceeds the range of dyno testing.
Another complication. Some diesel emissions control devices do not function in a continuous second-by-second manner. Rather they store up some pollutants and then burn them off in discreet chunks. If the store-up phase is longer than the regulation dyno-test, and the dyno-test happens to miss the less-efficient burn-off phase, then are you cheating? What if the maximum long-term efficiency of the system (in terms of extra fuel used during burn-off vs quantity of pollution burned off) is maximized by long store-up cycles that are genuinely longer than the regulation test cycle?
Someone in this thread reminded us that the law is an ass. It seems clear the law is simply not capable of dealing with the full spectrum of modern emission control systems. Taking legal code that was written in the 70's and applying it in shrill tones to modern cars does not seem constructive. It can be applied to simple cases like VW, but beyond that it's pretty difficult. Think hybrid cars where (according to an earlier poster) they can start a dyno test with full batteries and end with empty. Cheating, yes? Or no?
I assume the purpose of regulating car pollution is to reduce pollution because it's a negative externality. Haven't we reached the point where it's technically feasible to put regulated sensor systems on individual cars and then bill owners for emitting pollution? The charge per unit of pollution would simply be adjusted until the resulting total pollution (from all cars) hit a desired annual limit as determined by the government. Instead of yelling louder about the exact test proceedure the government should follow to group-certify vast numbers of cars, couldn't we just let the owners drive them and pay for pollution as actually emitted? This incentivices car-owners and car-makers in a much more direct way than the current certification system.
IF it was technically feasible to measure the pollution from individual real-world cars then is there any functional reason not to do it that way?
Thanks for reading my Ciro-esque post.
Admittedly Ciro is more interesting.
.
