[electrons]

It is not a pure software fix, since, in the event of a detected fault, it is intended to alert the owner....

An alert is from software. Still all pure software. ..... char faultmessage[] = "Hot Times Tonight"; .... No hardware changes from what they've said so far.

They think a battery will give many hours warning, sitting in a driveway, triggering faults from a running embedded computer, waiting to alert the driver later on to take it in for repairs immediately or get marshmallows ready for roasting.

 

[desiv]

They think a battery will give many hours warning, sitting in a driveway, triggering faults from a running embedded computer, waiting to alert the driver later on to take it in for repairs immediately or get marshmallows ready for roasting.

Possible they believe their software "fix" will detect the issue weeks/months before it reaches the point it could catch fire.
We really don't know, but that could be the other end of the spectrum.

 

[shrox]

I'd certainly make my decision based on speculation rather than facts...

 

[desiv]

I'd certainly make my decision based on speculation rather than facts...

Good thing no one really has to make any decisions now, because there are few facts. ;-)

Perhaps we will have some by April.

The good thing is that leaves plenty of time for speculation, which can be fun, even if it isn't something you should decide things on... ;-)

 

[electrons]

The more info GM provides, the more people can decide to ask for a GM buyback, or to get a used Bolt after the recall supposedly fixes the problem. Confidence in the solution is key for GM.

Where it concerns battery fires, GM & LG Chem should be forced to share solution info with us and competitors.

Kind of like how the NTSB produces aircraft accident reports & goes into a lot of internal detail about control laws, etc., that contributes to the dangerous conditions.
I'm all for IP, yet fire safety is serious.

 

[sundog]

Perhaps we will have some by April.

Since it comes from the same people as the "ASAP after Jan. 1" communication, I'm thinking "in April" is code for May 31.

 

[desiv]

Since it comes from the same people as the "ASAP after Jan. 1" communication, I'm thinking "in April" is code for May 31.

Could be..
I have Hilltop Reserve on, so I am OK with that.
If it takes a bit longer to get the firmware right, then take that time... ;-)

 

[GJETSON]

Since it comes from the same people as the "ASAP after Jan. 1" communication, I'm thinking "in April" is code for May 31.

Since we haven't been inconvenienced so far, I won't be volunteering to be one of the early testers of this patch. If it sounds good after a month or two of reports, I will schedule an appointment.

 

[Colt Hero]

An alert is from software. Still all pure software. ..... char faultmessage[] = "Hot Times Tonight"; .... No hardware changes from what they've said so far.

Right! And when you think about that you realize that the information to write this logic was there all the time - right from the beginning ... so why wasn't it already in there ... or why wasn't it developed sometime over the course of the last 4 years?

 

[Colt Hero]

If we assume each fire was due to one bad cell, that means we've had four failures out of twenty million cells in a population with an average age of two years. So a failure rate of 11.3E-12 per hour. If GM has 100 battery packs in extended testing and they're the same average age as the general population (assume GM sample each batch that arrives from LG) then the chance that one of them develops this issue is about 0.6%. So I'm not surprised GM didn't catch this in advance. The question remains: can they reliably detect a cell with dendrite growth with the diagnostics built into the battery pack?

All that cell math and "4 fires in 68,000 units" will become irrelevant when it comes out that these fires were caused by a failed Separator - a failure mode GM surely knew about, but failed to write proper detection software for (to avoid the fires).

And inside the walls of GM right now, while they're scrambling to update their software, are probably engineers thinking "boy ... we were lucky it was only 4 in 68,000"!

 

[sundog]

In the style of "thereifixedit.com", the fix will be: permanent HTR, plus delete of the DCFC option (but don't worry... you'll get the full $700 option cost refunded!).

 

[RPC]

All that cell math and "4 fires in 68,000 units" will become irrelevant when it comes out that these fires were caused by a failed Separator - a failure mode GM surely knew about, but failed to write proper detection software for (to avoid the fires).

And inside the walls of GM right now, while they're scrambling to update their software, are probably engineers thinking "boy ... we were lucky it was only 4 in 68,000"!

Perhaps, but my comment was directed at the "why didn't this show up in testing" crowd. In order to have even a 50/50 chance of uncovering this, GM would have had to have something like 10,000 packs in the lab being cycled full time. If you need that level of certainty, the only realistic choice is to not release the product.
And yes, the engineers probably ARE saying just that - it looks like these cells have a lower failure rate than a piece of wire.

 

[TheLondonBroiler]

All that cell math and "4 fires in 68,000 units" will become irrelevant when it comes out that these fires were caused by a failed Separator - a failure mode GM surely knew about, but failed to write proper detection software for (to avoid the fires).

And inside the walls of GM right now, while they're scrambling to update their software, are probably engineers thinking "boy ... we were lucky it was only 4 in 68,000"!

There have been 8 reported Bolt/Ampera-e battery fires:
3 - 2019
1 - 2018
2 - 2017
1 - ? Opel Ampera-e - Germany
1 - ? Chevrolet Bolt - Ukraine

 

[Sean Nelson]

The fix Boeing is adopting has been tested up the ying-yang to be safe. What started as one of the most reliable families of planes in history has become even more reliable.

The family has a stellar track record that's been ruined by the Max, because it was a totally botched design. It's asinine that they'd take their two angle of attack sensors, essential to control a flight-critical system, and use only one of them, alternating which one is active with each flight. They "fixed" the software to use both and to only activate the MCAS system to push the nose down if both agreed. The fact that it wasn't done this way from the start beggars the imagination. It would be like designing a twin engine aircraft and deciding to only use one engine per flight because it probably won't fail, right?

The Bolt, by comparison, even with its battery issues, seems far better thought out.

 

[MisterJJ]

The fact that it wasn't done this way from the start beggars the imagination.

I've seen this trend in engineering where there are so many procedures and tools used for design and validation that nobody sees the forest for the trees. Instead of experienced engineers teaching and showing new engineers how to think through potential consequences, new engineers are forced to follow a pre-packaged design system which is expected to prevent errors like this.

Don't get me wrong about these design tools though. They can be very useful in exposing defects in design that would not otherwise be found. But they've become a crutch that is relied on far too heavily. I'd love to see the root cause analysis on this design failure.

 

[Colt Hero]

I'd love to see the root cause analysis on this design failure.

I think the "root cause" was: knee-jerk decision-making, reacting to AirBus getting ahead of them and stealing their orders, going with a fast-tracked, short-cut kludged design, retrofitting oversized engines onto an aircraft they weren't designed for. Then trying to fix all the problems this caused through software (which they ultimately botched).

In retrospect, I'm sure they came to the conclusion that they should've stuck with their original plan: to properly design and build a new plane from the "ground up".

 

[desiv]

I think the "root cause" was:

That's the great thing about a real root cause....
It isn't what someone thinks it might be. ;-)
Yes, I realize speculating without real info is fun, but the point of a root cause is to know without guesses/thoughts.

So I'm looking forward to that as well.. ;-)

 

[electrons]

See this? The Kona battery replacements will happen in the U.S. too, not just Korea as had been previously reported.

Hyundai Motor to replace battery systems in $900 million electric car recall

Hyundai Motor Co will replace battery systems in some 82,000 electric vehicles globally due to fire risks - a costly $900 million recall that lays bare the thorny issue of how car and battery makers split the bill when problems arise.

www.reuters.com

Reading that article, LG Chem says it's not their China-based factory that caused the fires, it's Hyundai's failure (refusal) to apply LG’s "suggestions" for fast-charging logic in the battery management system (BMS), adding the battery cell should not be seen as the direct cause of the fire risks. .... Which appears to be GM's blame vector! GM just wants to fix the Bolt issue with clever-er BMS fault detection algorithms, not battery replacments.

 

[zendroid1]

Does this influence Chevy's decision at all? I'd say not likely.

 

[KiwiME]

... not just Korea as had been previously reported.

That's because it's more recent news, not a misinterpretation of the former report. And it's global.

I can think of three Kona incidents that happened while or soon after AC charging, two cars were brand new, one still on the lot. It's still unclear if packs are being replaced because of defective or damaged cells, and/or because the BMS is too highly-integrated to repair in-situ.

Konas owners such as myself are waiting to hear GM's response.