Phoenix Range Test Results, September 15, 2012

[Stoaty]

And so, it appears you may have misrepresented the range loss of all the LEAFs tested, by exaggerating the percentages of range lose by about that same 5%. And since you apparently did not collect any data that would allow us to try to estimate actual battery capacity of any of the test cars:

So, the only thing I can conclude from the methodology and results of of this range test, its that eleven of the twelve LEAFs (~ten of which were selected for testing because the owners reported unusually high loss of range) have about 84% to 100% (with a large and unknown level of uncertainty) of what Nissan states (in your single source cited) is "estimated" new LEAF range.

Would it have been better to have a Leaf with 281 Gids at full charge? Of course. However, one thing is clear from Tony's data: the percent Gids is lower than the percent of a "New Leaf Range" (84 miles) in every single case. It is reasonable to infer that a Leaf with 100% Gids would have at least 100% of the "New Leaf Range"--and likely more. My guess is that a Leaf with 100% Gids would travel farther than 84 miles, such that the range loss is probably greater than calculated from the data. If a new Leaf can be found and tested on the same course in similar conditions, I predict that this will be the case.

 

[thankyouOB]

i looked at the data.
the range does not seem all that bad for these cars, considering that they are supposed to be paradigms of troubled Leafs.

what do we conclude from this?

 

[WetEV]

="TonyWilliams"

Those (m/kWh) numbers were ALL OVER THE PLACE. Not only is that data not needed, you can't prove much with bad data.

It is irresponsible for you to refuse to disclose this "bad data".

Please release the m/kWh numbers.

There is no "bad data". Data can show things other than what is expected, can be not useful, can be hard to interpret, but is never "bad".

The information from the car about the battery is non-ideal. This makes battery loss estimation a lot harder than just plugging in a CAN bus monitor or looking at the capacity bars.

Note that while the battery isn't warrantied, the rest of the car is. That would include the instrumentation. Understand?

Nissan might be as puzzled about parts of this as the rest of us are.

 

[edatoakrun]

And so, it appears you may have misrepresented the range loss of all the LEAFs tested, by exaggerating the percentages of range lose by about that same 5%. And since you apparently did not collect any data that would allow us to try to estimate actual battery capacity of any of the test cars:

So, the only thing I can conclude from the methodology and results of of this range test, its that eleven of the twelve LEAFs (~ten of which were selected for testing because the owners reported unusually high loss of range) have about 84% to 100% (with a large and unknown level of uncertainty) of what Nissan states (in your single source cited) is "estimated" new LEAF range.

Would it have been better to have a Leaf with 281 Gids at full charge? Of course. However, one thing is clear from Tony's data: the percent Gids is lower than the percent of a "New Leaf Range" (84 miles) in every single case. It is reasonable to infer that a Leaf with 100% Gids would have at least 100% of the "New Leaf Range"--and likely more. My guess is that a Leaf with 100% Gids would travel farther than 84 miles, such that the range loss is probably greater than calculated from the data. If a new Leaf can be found and tested on the same course in similar conditions, I predict that this will be the case.

And your guess, may be better than mine.

Since all my level freeway speed driving has been on long trips requiring slow L2 charging, I have never had the luxury of the extra time required to test my LEAFs full range at this speed myself.

I have always thought the range estimates I got from Nissan before I took delivery, and those posted by actual road tests, and later assimilated by Tony into the range chart, were close to correct.

And I would correlate both those sources of estimates, with (very roughly) between ~76 and 80 miles, under the Phoenix test conditions.

Even if you scour the country find a "new" LEAF with the magic gid count you seek (how often do you see 281, at Phoenix range test temperatures?) what evidence would you have that there is not a significant variation in available battery capacity between "new" LEAFs, and that that single LEAF was not atypical?

As an example, I would guess, that "Blue 534" from the test probably would have done 84 miles when new.

 

[WetEV]

Would it have been better to have a Leaf with 281 Gids at full charge?

I don't trust that "Gids" or capacity bars are accurate. Is there a reason to?

Two cars traveled 79 miles in the test. One had "85%" based on Gids and 12 bars capacity, probably also based on Gids, the other had "75%" and 10 bars.

 

[Stoaty]

I don't trust that "Gids" or capacity bars are accurate. Is there a reason to?

I agree, they are not accurate. However, the values are not random. We see good evidence that there is a systematic bias that under reports the available range. Thus it is reasonable to infer that a Leaf with 100% Gids would have at least 100% of the "New Leaf Range". I can't state it any more plainly than that. Testing a Leaf with 281 Gids is the only way to prove this inference correct.

 

[TickTock]

Looks like three of the cars dropped a few gids between the time Tony's crew took their data and when I came and repeated at 5am. Was only to capture post drive data for three cars since I was one of the last drivers to return. :-(

 

[edatoakrun]

I want to reiterate why I think anyone undertaking a range capacity test should record and report all of the m/kWh results from their tests.

It is entirely possible that the gid count itself is the (or, at least one of several) sources of inaccurate kWh use data, the very problem that Nissan is now evidently facing.

As posted a couple months ago:

Ever since TickTock fist (sp) suggested the topic of gid variability, the implications of his observations have been setting in.

"TickTock"

1 gid *mostly* equals 80Wh

http://www.mynissanleaf.com/viewtopic.php?f=44&t=9689" onclick="window.open(this.href);return false;

If “gids” do reflect a variable amount of Wh, and they are the values used by the LEAF to calculate kWh use, then the capacity bar displays, dash and nav screen displays of m/kWh, as well as the Carwings calculations based on these same “gid” values, might be expected to be incorrect as well.

I now believe that this quite possibly could be be the case....

http://www.mynissanleaf.com/viewtopic.php?f=31&t=9064&start=20" onclick="window.open(this.href);return false;

 

[turbo2ltr]

So the one thing we know for certain is that we don't know anything for certain. lol.

Don't get me wrong, I understand and respect the pursuit of data. The test has shown a trend that BMS's that think they are connected to batteries with lost capacity trigger turtle after driving less miles. But that's about it.

I'm not of the camp that thinks this is an instrumentation error, but so far no one has proven it isn't. Unfortunately no one will without diving deep into the BMS code, of which I don't think anyone has access to.

Bottom line: Batteries are complex. SOC is a complex calculation. Battery charging is complex. Battery chemestries are complex. There are a lot of variables. More than any test could account for. Not to mention there is this blackbox of a BMS that nobody has a clue how it operates internally. Without this information, I'm afraid we are just spinning our wheels.

I hate to be a naysayer, and maybe if I had bought, I'd be singing a different tune. But I respect Tony and his minions that helped out with this test. It was a HUGE undertaking.

 

[kolmstead]

i looked at the data.
the range does not seem all that bad for these cars, considering that they are supposed to be paradigms of troubled Leafs.

what do we conclude from this?

That for freeway driving, with A/C on and going no lower than LBW, the 4-bar car has maybe 45 miles of range.

-Karl

 

[turbo2ltr]

I would love to compare mfgr date of the battery modules. I just can't believe the 2012s are so low. It would also explain the fact that so many people lost a bar within a few months, somewhat regardless of the length of ownership, miles driven, and charging behavior. (all of which I believe also play a factor)

 

[WetEV]

I don't trust that "Gids" or capacity bars are accurate. Is there a reason to?

I agree, they are not accurate. However, the values are not random. We see good evidence that there is a systematic bias that under reports the available range. Thus it is reasonable to infer that a Leaf with 100% Gids would have at least 100% of the "New Leaf Range". I can't state it any more plainly than that. Testing a Leaf with 281 Gids is the only way to prove this inference correct.

The correlation between range and Gids in this test is ~80% (note: sample size too small). So yes, the values are probably not random, but that can't be completely excluded. There is also likely one or more other factors, the exact type of which is unknown. We see underestimated available range, but the odds that Gid count might overestimate available range are not negligible from the data in this test as the sample size isn't large enough. Testing a new Leaf that also has 280+ Gids would give a single point to determination of the "New Leaf Range", and ideally enough such tests would be run to get a sample size large enough to draw inferences that had statistical meaning. This test isn't large enough, and testing a single new Leaf also isn't large enough to be statistically meaningful.

Sorry, I'm an engineer. It is easy read too much out of too little data, and see things that don't hold up when more data arrives. It is important to get enough data, and to realize that the situation is often more complex than anyone might first think it is.

 

[shrink]

Sorry, I'm an engineer. It is easy read too much out of too little data, and see things that don't hold up when more data arrives. It is important to get enough data, and to realize that the situation is often more complex than anyone might first think it is.

While we all acknowledge the N is small, please keep in mind the logistical challenges of running this test. Tony cannot be thanked enough. A lot of attention was also given to safety and picking up the turtled cars as fast as possible. While far from perfect lab conditions, it was an excellent sample of real world driving.

That said, this small N was also double the N of Nissan's own Casa Grande 6.

 

[shrink]

Oh and for all those wondering, my car was Blue 842, the 2012 with 2500 miles on it.

I drove White530 and per the dash, I got 4.0 miles/kWh.

 

[WetEV]

Sorry, I'm an engineer. It is easy read too much out of too little data, and see things that don't hold up when more data arrives. It is important to get enough data, and to realize that the situation is often more complex than anyone might first think it is.

While we all acknowledge the N is small, please keep in mind the logistical challenges of running this test. Tony cannot be thanked enough. A lot of attention was also given to safety and picking up the turtled cars as fast as possible. While far from perfect lab conditions, it was an excellent sample of real world driving.

+1

That said, this small N was also double the N of Nissan's own Casa Grande 6.

Hmm... that raises all sorts of interesting questions.

 

[TonyWilliams]

we shouldn't expect that they will bother responding to the AZ Tempe test results at all, do we?

You can all fly up to San Francisco Crissy Field on Sept 23 (it should be gorgeous with the Golden Gate in the background) and meet the decision makers yourself. They will be there for National Plug In Day.

Andy Palmer (<<<-- probably not) and Mark Perry will be there, so you can ask them directly.

 

[evnow]

My suggestion is to release ALL the collected data.

As I have written in the past GID is not linear i.e. at the top & low end a GID has less energy than in the middle. That is the reason I stopped recording gids after every trip - and didn't post any definite conclusion from collected data.

The fact that 2 cars with 12 & 10 bars can have the same range points to faulty metering in the cars (even though the correlation of gid/bars to miles is high). Atleast it points to unreliable metering. That is the first thing Nissan needs to fix.

 

[Stoaty]

As I have written in the past GID is not linear i.e. at the top & low end a GID has less energy than in the middle.

In my two range tests, I found that Gids at the low end (15-25%) had more energy than the middle.

 

[edatoakrun]

Oh and for all those wondering, my car was Blue 842, the 2012 with 2500 miles on it.

I drove White530 and per the dash, I got 4.0 miles/kWh.

Thanks.

Using the reported gid count (assuming 85% was representative of the actual gid report error rate) to test for the same error rate on the nav screen (assuming you would have seen a reported 4.1 m/kWh there) I believe you would get a calculated ~22.8 kWh of available capacity for that car's battery pack, on that day.

The problem is, It was not a new battery, so you'd have to reduce that ~22.8 kWh figure by whatever was the correct % loss of capacity this battery had, from new to find the actual gid error % rate and derive the actual kWh capacity from that.

And since no one (apparently) tried to calculate the battery capacities of the test LEAFs, from any data source independent from the (erroneous) gid count, the actual gid error rate is now not calculable.

There is no real way I can see to verify this attempt at backwards analysis, other than to see how closely the unreported m/kWh report error rates from the other cars, match (or do not) the gid report error rates from the other test cars.

If consistent correlation is observed, we might be able calculate the actual available battery capacity, if we actually knew how many m/kWh these cars actually achieved, on the test.

But of course, if we actually knew the m/kWh, we would know the kWh capacity already...

Looks to me that if you want to know your LEAFs actual battery capacity, a meter or (accurate charge rate) and the charge efficiency, will be required.

If you want to know the range, be sure to charge as well as drive it in controlled range test, and you can ignore the capacity bars, and gid count.

 

[Volusiano]

we shouldn't expect that they will bother responding to the AZ Tempe test results at all, do we?

You can all fly up to San Francisco Crissy Field on Sept 23 (it should be gorgeous with the Golden Gate in the background) and meet the decision makers yourself. They will be there for National Plug In Day.

Andy Palmer and Mark Perry will be there, so you can ask them directly.

Any San Francisco local owner(s) volunteering to attend and question Nissan execs about what they think of the Tempe, AZ test results?

I suspect they would dismissively say they never saw it, or that they don't care to see it.

I think it'd probably be more effective if somebody can urge the media to ask them this question, in front of the cameras.