Car and Driver: These Tests Failed You: Why Is the EPA So Bad at Estimating Hybrid Fuel Economy?

Reading that piece, it looks like Ford may have just made up it's additional EPA numbers, rather than run the tests.

 

If the headline had said *Ford hybrid fuel economy* I would understand.
Just how remarkable is Toyota's HSD ? 15 years after HSD was available for tear-down STILL no other manufacturer can replicate, let alone beat it.

If the article would mention what the fudge factors are we could judge whether the problem is in fact the interpolating. I'm still more perplexed by the apparent inability of people to replicate EPA using drive routines that lead to > EPA in a Prius.

 

The Car and Driver article appears to be thorough and well balanced.
Much more so than I expected -- shame on me.

It would appear that Ford didn't actually cheat, rather it used the most
advantageous post actual test EPA permitted statistical fudge factor.
However, there is the possibility (probability?) that Ford's initial actual
test figures were themselves influenced by specifically programming the
hybrid control algorithm's to the test requirements, as opposed to "real
world" conditions.

IIRC, there have been similar speculations/complaints about the generally
difficult to achieve/replicate Hyundai/Kia MPG figures. It remains to be
seen if the same MPG discrepancies will exist for VW's new turbo hybrid.

 

No shame Rokeby -- starting the article by mentioning that a lawsuit is ongoing is pure crap. As if that means *anything* in sue happy America.

 

It seems like a really hard problem. Capturing the efficiency of a vehicle in a single number would certainly useful, numbers can be compared and sorted, etc. But I don't know if it's possible to do in a way that satisfies everyone.

One possibility would be to make the tests very intense so they represent the worst case for fuel economy. Then no one would be disappointed by their post-purchase experiences. But that would probably fail to capture the potential of a carefully driven hybrid.

Does anyone know what a Prius scores on the older CAFE fuel economy tests?

 

Gen 2 was 60 city/51 highway

 

No reason why the the CAFE numbers used for regulation compliance couldn't be printed on the sticker along with a worst case one, which would be the only one allowed in advertizing.

 

...if they are improving the EPA test, I would like to see an E10 cycle too so we can get an idea for ethanol impact.

 

I agree that its difficult to see an easily implemented way around the problem
if it has to do with control algorithms maximized for the standardized test.
Sort of the engineering equivalent to "teaching to the test" in the educational
area. Students pass the test but are not really sufficiently educated or intellectually
flexible enough to pass onto more challenging material or function fully out in
the real world. (Yes, I know, a wholly different subject for disciusion. Please,
should not be used to hijack this thread.)

A proposal, probably simplistic. Have 3 to 5 variations of the initial must-actually-
be-done before permitted fudge factors are applied test sequences. Which test is
used is randomly chosen. Might actually just be changes to order or length of the
different operating conditions now used. Of course, the different tests would have
to output the essentially equivalent numbers... to what, +/- 5% or less.?

Permitted fudge factors also need to be re-evaluated.

Yes, I know an administrative nightmare, but my best idea.

 

I think the only change that is needed is to add this text to the EPA label:

"The results are based on non-agressive driving, gentle braking, and speeds less than 70 mph 90% of the time. Driving like a moron will dramatically affect the results YOU obtain with this car."

 

It is a very good analysis and I believe it is correct.
The best thing for EPA to do is to enforce all five tests for hybrids, that should make the numbers closer to real world.

 

That ought to be easy to compute. For the 1985-2007 EPA scale, find the CAFE numbers (I don't have the link at epa.gov handy, but Cwerdna has posted it many times), and multiply the City test by 0.90, the Highway test by 0.78.

For the Gen3 liftback, I think that will produce about 63 or 64 city, 55 highway. But we need that CAFE list to be more certain.

 

The only problem with this article is the 'broad brush' of using the worst offending vehicle as if all hybrids suffer from the same abuse. Given the Prius is the largest part of the market, this report would have been more accurate if it compared the Prius 'real world' numbers versus the Ford.

I had suspected vendors were allowed to 'fudge' some of the roll-down, numbers when I compared the actual and dyno coefficients for Ford. Toyota has also 'adjusted' their coefficients including . . . a 'negative' coefficient.

Bob Wilson

 

Our Congresscritters would never allow that because their agribusiness masters would cut off their allowance.

 

But I believe these are the CAFE numbers I mentioned. Thanks for repeating the link.

For the 2013, that should produce Old-EPA-City = (71.64 unadjusted) * (0.90 city fudge factor) = 64.48 ==> 64 mpg on the 1985-2007 scale.

The Old-EPA-Highway = (69.44 unadjusted) * (0.78 highway fudge factor) = 54.16 ==> 54 mpg.

Note that on the New (2008+) EPA scale, the Combined figure is 49.5027, which only barely rounds up to the Monroney EPA label of 50. Remove any single mpg improvement, no matter how small, and the EPA label would suddenly drop to 49 mpg, a serious marketing failure.

 

How would you do that? Self reporting is highly suspect. You would then ask car and driver and consumer reports, which as you know get much worse results on their "real world" tests for prius, camry hybrid, etc. It does mention that it did relatively worse for the ford hybrids versus the toyota ones. To car and driver and consumer reports the Prius falls very short of its sticker, it just doesn't fall as far as the c-max.

To me the major takeaway is this -

I'd say they need to do both, and perhaps publish all 5 cycles, including revised high speed one, on their website.

I don't think that is what car and driver found. I would mail them and check. It appears that Ford used a fudge factor from the 2 cycle test, which removes the high speed and cold tests. We can expect that ford would do worse if they actually ran the 5 cycle test, but by allowing the fudge factors, they were allowed an inflated number. IMHO the city test and highway test don't look like anything we drive on at least in my part of the country in 2013. Fudge factors instead of actually running the tests exagerate the benefits of certain technologies.

 

My understanding is that the pre-2008 EPA score are different from the CAFE scores. Am I wrong?