Second Fire cause by impact with a metal object. Recall necessary?

[jerry33]

IMO to avoid the suspension drop would likely not solve the problem because to prevent the accident with the road debris you could need to raise the car another inch in a millisec. (Maybe some calculations would be needed)

I was thinking of a manual setting. Hitting road debris typically happens when there are lots of vehicles around you. In those situations the driver could (should) elect to not have the automatic lowering. Aerodynamics won't be affected that much because a stream of moving vehicles reduces aerodynamic drag. When the traffic is light, then the automatic lowering could be reengaged.

 

[FredTMC]

The only change to the car that makes sense to me is a software patch that lets us choose the high speed ride height. If you want, you could select standard height if you are going to be in heavy traffic and have no chance to swerve around an obstruction. If you are going to be in open-road conditions where you can see well ahead, you could choose to use the lower setting.

I agree completely and proposed the same earlier. I love the idea. It's my belief that the current "low" height setting should be optional and driver selectable.

 

[SteveG3]

Fred, Thumper, very good idea.

 

[32no]

This Tesla Model S in this video didn't catch fire after running over some unknown object that caused the car to jump about 3 feet in the air. I think that the other accidents are just freak accidents.

Accident big rig on 101 NB at Woodside Rd. on 10/18/2013 around 10:50AM - YouTube

 

[brianman]

Model S ... certainly catches fire more easily than a horse, maybe that's what people are comparing it to. :wink:

Public plea: If you want this tested, please leave it to Mythbusters to do so.

 

[azure1979]

How about attaching liquid metal at front part of aluminum plate amour. The liquid metal is 10 times stronger than aluminum in the same thickness. I think just small liquid metal is enough to protect battery from metal debris.

 

[JRP3]

I assume you're talking about this: http://liquidmetal.com/properties/the-science/

Very expensive as I remember, and difficult to produce in larger sizes.

 

[Jhall118]

Stop the madness.

As a small time shareholder and someone who is looking at buying a Model S RIGHT now, I think Tesla would be silly to modify anything. There will always be accidents- that's a reality of driving. The fact that the Model S is designed to completely protect that passenger, and give notice for them to pull over, is the real point here. If they waste the time/money/resources to reinforce the pack I would honestly be dissapointed in the company. No matter how well you reinforce it, accidents will still happen, and may cause the car to ignite again. This would lead to the media making an even BIGGER deal, about such a non-issue.

It saddens me that people on TMC are calling for a recall or redesign. I don't think it will have a long term effect on demand, and any short term effects will be pointless, because they are supply constrained.

 

[SwedishAdvocate]

Cross-posting from the humongous other thread since this seems to be a parallel discussion:

There is a car fire every 96 seconds in the US. Some with fatalities or severe injuries. We had only three, every time driver walked away unscathed. Please let us not make a "Mountain out of a Mole":

National Vehicle Fire Statistics | Chandler Law Group

Again, what happens with other vehicles is irrelevant, perception has already made this a mountain.

Wasn’t that also the case after the first fire?

It seems that was pretty easily fixed with a blog post on the Tesla website and a couple of media appearances by Elon.

As I understand it Apple also faced a perception-problem with the Iphone 4 ”antenna-gate”. Enter Steve Jobs:

Steve Jobs at the Antennagate Press Conference (2010) - YouTube

And as I understand it the Iphone 4 sales were just fine (spent 2 mins. searching for some numbers, but couldn’t find any…)

Are we really sure this perception problem is unfixable?

There was also a Q&A with Jobs, Cook & Mansfield after the presentation that I link to above, but I can’t find a good video of that one…

Full disclosure: I have not voted in this poll.

 

[Jhall118]

I think Tesla made the perfect call. Let the customer speak. This was probably better than anything Elon could have said himself (what could he say that hasn't been said before?). It's incredibly important that everyone involved in accidents runs out and gets another one (versus not being able to run at all because they are dead). The guy was even able to grab some papers out of the glove box afterwards, and open the doors? That is incredible.

The bottom line: Tesla isn't suddenly going to have more supply than demand anyways. By the time Gen 3 rolls out, we will have REAL statistical data about the fires anyways, not data based on a relatively small sample size. Enjoy the fire sale, because I feel an announcement is coming "very soon" about giga factory (they said so in conference call for crying out loud!)

 

[JRP3]

The market may see the Giga factory as a huge expense, hence capital raise, hence dilution.

 

[brianman]

Stop the madness.

+1 (to the post as a whole)

 

[Chrisr400cc]

Tesla should probably do a recall and do something to make the pack tougher. 2 times debris punctures the pack in a month is not good. We can't just have Model S's driving down the highway running over stuff and catching on fire. It's just not good. A third one could happen anytime. Then what? I think it would be worth it to recall the 19,000(?) model S's and put something on them to protect the battery better. It would have to be cheaper then having the company drop a billion dollars in valuation every time there is a fire. Hope Elon comes up with something good this week

 

[drinkerofkoolaid]

The market may see the Giga factory as a huge expense, hence capital raise, hence dilution.

Who said Tesla would be the one funding the Giga factory?

 

[Raffy.Roma]

Tesla should probably do a recall and do something to make the pack tougher. 2 times debris punctures the pack in a month is not good. We can't just have Model S's driving down the highway running over stuff and catching on fire. It's just not good. A third one could happen anytime. Then what? I think it would be worth it to recall the 19,000(?) model S's and put something on them to protect the battery better. It would have to be cheaper then having the company drop a billion dollars in valuation every time there is a fire. Hope Elon comes up with something good this week

Since actually it's not a matter of safety but of PR/image (because in all the fires people inside the cabin got out of the Model S safely) IMO a change to the inside of battery pack (chemistry, modules) should be done for future Tesla cars to prevent that they catch fire after hitting a road debris. This change should be done not so much to improve the safety of the Model S, that is already very safe, but to prevent that in case of fire the car is totalled.

 

[Grendal]

Tesla should probably do a recall and do something to make the pack tougher. 2 times debris punctures the pack in a month is not good. We can't just have Model S's driving down the highway running over stuff and catching on fire. It's just not good. A third one could happen anytime. Then what? I think it would be worth it to recall the 19,000(?) model S's and put something on them to protect the battery better. It would have to be cheaper then having the company drop a billion dollars in valuation every time there is a fire. Hope Elon comes up with something good this week

What if the battery pack isn't the problem? There is a lightweight plastic aerodynamic covering in front of the pack. If the objects are penetrating that weak plastic then you can armor the pack until it is as strong as a battle tank and you would still be in the same position. Before anything happens, Tesla needs to determine where and what is the weak spot that allows the pack to combust.

 

[JRP3]

Who said Tesla would be the one funding the Giga factory?

I did say "may". I do think a partnership or separate company is more likely, but depending on how it's structured Tesla may still be investing significant amounts of money.

 

[drinkerofkoolaid]

I did say "may". I do think a partnership or separate company is more likely, but depending on how it's structured Tesla may still be investing significant amounts of money.

Possibly. Here are a few links I think you might find interesting.

EV Battery Reuse Research Partnership Between BMW Vattenfall Announced | CleanTechnica
Giant A123 Electric Car Battery Plant Opens in Michigan | PluginCars.com
More details on Ford partners lithium-ion battery plant in Michigan
GE Expands New York Advanced Battery Plant ecomagination
Hollande hails new lithium battery plant - FRANCE - FRANCE 24
http://www.jsonline.com/business/42963372.html
Panasonic, Call2Recycle Partner on Battery Recycli - RT - Recycling Today

 

[smorgasbord]

So how many months and how many miles went by without a single incident? Two in one month would mean something if it were the first month the cars were on the road.

I'm not an expert in statistics, but I don't think this is quite true. If you have a "fair" coin and toss it 99 times and 99 times it comes up heads, the odds are still 50% that the next toss will come up tails. So that no fires happened for the first 10-12 months and now happened twice the same month doesn't mean the recent occurrences are freaks - it could be the initial stretch was a freak.

And then you have to factor in the known variables. In the first months there were only a small number of cars on the road - as each month goes by not only have more cars been put on the road, but the rate at which they're entering the roads is increasing. There could be other factors as well - as drivers get used to their Teslas and overcome range anxiety, and as more chargers and super chargers come online, more people are taking more trips on roads they don't normally travel.

The Model S design is akin to putting a small gas tank in the chassis of the car. I don't think any ICE car company does that. Yes, there's a 1/4inch of steel helping protect it, but from the first accident we know that steel armor can be penetrated by running over something, and the latest appears (yes, I'm jumping the gun here) to be similar. Perhaps there's something about speed and leverage that creates the enormous vertical forces needed to reach the battery. It'll be interesting to read the detailed mechanics of what happened in the latest fire.

Perhaps someone with more statistics knowledge than myself can weigh in here. I do remember reading this: The Three Door Puzzle - A Probability Question And Answer and not believing it, so perhaps I should run the example with some friends. It says that if you get to choose 1 door out of 3 and then Monty shows you one of the other doors and asks if you want to change you choice, statistics says you should, even though it doesn't feel like it would help.This isn't applicable to the Model S fires, but my point is that statistics aren't always obvious.

 

[Mario Kadastik]

Perhaps someone with more statistics knowledge than myself can weigh in here. I do remember reading this: The Three Door Puzzle - A Probability Question And Answer and not believing it, so perhaps I should run the example with some friends. It says that if you get to choose 1 door out of 3 and then Monty shows you one of the other doors and asks if you want to change you choice, statistics says you should, even though it doesn't feel like it would help.This isn't applicable to the Model S fires, but my point is that statistics aren't always obvious.

This is an example of Bayesian prior knowledge entering the system. At start you designate 1/3 probability to each door and that initially remains unchanged. If you pick one door your probability of getting it right is 33% while the probability that the win is in other doors is 67%. This does not change when the show host opens one of the doors because the host had prior knowledge that means that he knew 100% where the win is. Now considering that one door was opened proving the win wasn't there the 67% assigned to not-your-pick is all transferred to the other door you didn't pick. So you have 33% chance of winning when you remain with your initial door and 67% chance of winning with the swap. The main reason this works is because the host introduced added knowledge to the system.

With regard to statistics on the Model S fires if an event is probabilistic in happening X times per Y driven miles (the only decent statistic that in this case works, even better would be per driven highway mile due to higher speeds and lowered suspension), then the event count over an integrated time period is a random variable with Poisson statistics. At high enough event count the Poisson statistics converges on the normal distribution that most people use daily for statistics. However at low counts the Poisson probabilities should be used and it's an exercise I did on ca page 47 of the fire thread. If the average car has 1 fire every 20M miles driven, then at 100M miles the expected event count is 5. Even if you discount for the age of the car etc and reduce it (I think way too conservatively) to 1 event expected, then the probability to observe 3 or more events is 8%. That's hardly significant (we in HEP at least don't consider anything significant that is below 3 sigma or about 99.7% or in this case the inverse which means that the hypothesis that this is normal would be excluded if the probability of observing 3 events would be < 0.3%). If we keep the expected number of events at 1 (too low I think considering average statistics, but let it be), then to reach < 0.3% probability we'd need about 5 events (that's 0.36% so technically we'd need 6 events, but it's close enough).

If we take the average of the two (1 as very conservative, 5 as the statistical average across all cars) i.e. 3 events, then our current occurrence is 57% probable. To get to 0.3% we'd need >9 fires to make this significant.

Anyone can do these exercises with a simple poisson calculator. They take two arguments: expected occurrence rate and observation (i.e. random variable). Just google for poisson probability calculator.