I don't know why I didn't think of this before I read CapitalistOppressors excellent recap:
I think it probably doesn't count range that strictly. I believe it rather goes into power limit mode due to low voltage. When you pull lots of current from a battery, you will see progressively larger voltage dips as it drains. I bet the car limits power the first instant it sees the voltage briefly dip below some threshold. Resting voltage can be much higher, but the battery becomes "softer" as it depletes.
This has happened with every battery I have drained aggressively, from RC cars and planes to my NiCd EV. Li-ion, NiMH, NiCd and lead acid all behave much the same, and it's very pronounced in my EV. When it's down to 30% and I floor it, I can see the headlights dim slightly and hear the ventilator fan slow down a little. NiCds tolerate this reasonably well (should be avoided, I never do it anymore), lead really dislikes it and li-ion needs active protection to prevent that it ever happens.
Try again with a full charge, I'm about 90% sure this is what happened.
If this is the reason, then I think the car will perform at least down to 50% SoC. Tesla might also be able to tune the behaviour with a software update. It would be possible to limit power in such a way that voltage stays a little above the threshold, instead of waiting for it to exceed it and then cut power by 50%.
I wont disagree with you, because I don't know that much about how batteries operate on an electric car (or in general for that matter). My understanding is that Li-ion tends to maintain constant performance until it reaches a very low SOC, at which point performance falls off a cliff. But that understanding is based on information I got at Home Depot when researching power tools
However, I am comfortable in troubleshooting complex issues with less than adequate information. That's how I make money after all.
So here's the data -
Cinergi reports that power was cut when he had ~30 miles of range. His photo shows 27 miles of projected, and 31 rated (I assume there was a slight delay between the time that Cinergi noticed the cutoff and when he was able to take the photo in a moving car. Call it a 0.1 second delay before his bionic video implants kicked in and captured the image). I've heard that Rated miles nowadays is starting out at ~245 with 100% SOC, which means this started when MSP hit ~13% SOC.
Cottonwood (like any normal non-bionic human) didn't take a photo of his dash when power cut out. But he was clear that he only had ~100 Rated miles when he started this testing. That means that the "problem" started for him when he had ~41% SOC, but he was also putting much more load on the car.
The key (to me at least), is that Cottonwood didn't have the problem every time he pushed the car's limits. He would complete the majority of a lap under load, and then performance would cut out. After a time at slower speeds he would try again, and again was able to push hard for a period of time before performance was cut.
Given the totality of the information, I feel it points to Tesla having implemented a hard cap at ~30 miles projected range to reduce power output and help folks get home in our electric infrastructure deprived world.
The hypothesis presumes that Cottonwood's
Projected miles was hovering right at the ~30 mile point, while his
Rated miles were ~100 as he reported. Under track conditions I think that makes sense.
Accelerate hard and the computer drives Projected miles down below the cap and cuts performance. Because there is such a wide divergance between Projected and Rated, slower speeds pull the Projected numbers upwards with a very high delta which rapidly puts the car above the cap again. Rinse, repeat, time to change the software.
The problem that I see (potentially due to ignorance?) with software looking at voltage levels as opposed to a simple range cap, is that Cinergi was capped at 13% SOC while Cottonwood was capped while at ~41%. And Cottonwood wasn't capped every time he pushed the car hard, but rather had to push hard for a period of time before hitting what appears to be an arbitrary cap that disappears on its own minutes later, allowing you to push hard again. I'd think that if voltage at a given SOC is a given, then once you reach that point you wouldn't have arbitrary periods of high performance and low performance.
I suppose I have had some experience with batteries getting a bit of a power boost from being allowed to rest for a minute (re Home Depot). But that was mostly with NiCAD batteries. One of the things I like about my Li-ion tools is they don't seem to do that. A straight range cap just seems to fit better to me because Tesla has an immediate incentive to discover that problem (getting low on range on an EV can be dangerous) and implementing a simple "fix" that would explain all of the observed behavior. Voltage looks to me like an engineering solution to an engineering problem that we aren't certain exists. So I'll put my money on Occams Razor.
- - - Updated - - -
Could be premature range extension, but then you'd have to assume that the Laguna 1:51 car had this stuff disabled, right? Or maybe it was topped off before the fast lap. If so, a requirement to have SOC very high to avoid limp mode in one or two laps essentially says you can't realistically do a fun track dy with the car at the limit.
Also, 1:51 implies plenty of time at full throttle. Front straight, turn 6?, climbing up to corkscrew, etc. Prob get over 100 mph four times a lap.
It would be great if CapO is right, because this software glitch could be fixed without impact on wear and tear (in contrast to raising thermal envelope limits, which Tesla may be loath to do without canceling warranty).
Actually, my hypothesis is that there simply isn't a problem until you get below ~30 miles of projected range. If you start off topped off you can go quite awhile before your projected range falls that far. The ReFuel time trials were 1 lap, but my recollection of the ReFuel videos is that they also had other events where the cars were all going head to head and completing multiple laps. In several videos (which seemed to last ~15 minutes or more) it was apparent that the Model S was lapping many of the competitors who were shooting video, which indicates it was operating at high performance for a fair number of laps.