DIY EV builder catches pack on fire, burns up car and house

[kiev]

[Edit: The forum posts were taken down and the links are dead ends, so i have removed them from here. Not sure why they did that]

A scary reminder from the DIY group. A fellow built an EV using 6 Tesla model S battery modules as his battery pack. Fortunately it happened during the day and he got out of the house with his wife and kid okay.


We probably take for granted the high reliability of our OEM battery monitoring system--not sure if i would trust my own BMS enough to charge an EV in the garage...

 

[mdbuilder]

Bottom balancing, no cell monitoring, LOLz :shock:

Some statistician should run some numbers, those packs had something like 2,700 individual cells. If one cell has say a 1 in 2000 chance of dying at any one time, and you have 2,700 cells in the pack, what are your chances on any given cycle? How much worse does it get after your packs have been mechanically assaulted by the high G rapid stop of a collision that made them available at the wrecking yard?

 

[Phximiev]

The comments indicate that he may have overcharged a cell similar to (and mentioning) the EVTV fire.

See: http://myimiev.com/forum/viewtopic.php?f=13&t=2900

For a homemade EV, lithium-iron-phosphate cells?

 

[mdbuilder]

Yes, the chemistry is much safer and 1P vs. 74P reduce to odds of cell failure by orders of magnitude.

 

[kiev]

Now i wish i had copied some of the posts since the thread was taken down.

Basically one of the 6 modules was at 18V while the rest were at 21, so a 3V delta for a group of 6s74p connected in series with the other modules.

My theory was that at some point in it's life that module's voltage was pulled down below 2.5 vpc or 15 volts, and then subsequently charged. During charging after such an event metal dendrites will form and grow which will puncture the plastic separator sheet between the anode and cathode, and will internally short circuit the cell. This is one of the ways by which these cells can fail and catch fire.

 

[Phximiev]

Why was the thread taken down?

I thought it to be completely relevant given other posts.

 

[JoeS]

...Basically one of the 6 modules was at 18V while the rest were at 21, so a 3V delta for a group of 6s74p connected in series with the other modules...

Although spooked by the fire, I'm kinda ignoring that discussion (but had been mildly following it on the DIY forum), as he had no monitoring or management system at the cell level, no overall OVP, no chargetime regimen, nor a backup timer to shut the charging down after a predetermined chargetime.

Perhaps that thread disappeared because of fire insurance claim concerns?

Drifting slightly off-topic, kiev, what piqued my interest is your 6S74P designation. I take that to mean six cells in series forming a string, with 74 such paralleled strings resulting in the battery. As an aside, such a configuration would result in the possibility of lots of individual cell imbalances.

I would have written it differently: 74P6S, meaning 74 cells in parallel forming a submodule, then six of those submodules in series forming the battery.

In looking at the radio control sites, they often refer to 3S2P as meaning three cells in series (forming a string), with the "2P" meaning two strings are paralleled.

What is the correct designation for six modules in series, each module containing 74 paralleled cells?

Going even further off-topic in this off-topic, in one of my Sparrows I'm running two paralleled strings, each string consisting of three Enerdel modules, with each module consisting of twelve elements in series, with two paralleled cells in each element. Umm, how should I designate that configuration?

 

[jray3]

Joe says "in one of my Sparrows I'm running two paralleled strings, each string consisting of three Enerdel modules, with each module consisting of twelve elements in series, with two paralleled cells in each element. Umm, how should I designate that configuration?"

I think that would be a 2P12S3S2P Pack...

Regarding the DIY Tesla salvage pack fire, I'm much more confident in the fire safety of large format LiFePO4 prismatics for DIY conversions, and seriously considered LEAF modules for the price, but am in wait-n-see mode on Tesla modules.

 

[kiev]

So are the Enerdel modules in series? i don't know how to designate your pack Joe, but the 6s74p is one tesla module, and there were 6 of these modules in series. Maybe it could be written as 74p6s, seems to be the same either way. 6s gives you the voltage, 74p gives you the capacity. The 74 cells in parallel are called bricks, sheets, or Cells by various vendors--so a module has 6 bricks in series, or 6 sheets in series, etc...

 

[mdbuilder]

In rc we refer to both configs S then P, the S gives the pack voltage overall. When you buy a large pack sometimes the cells are paralleled then joined in series inside the pack or module and sometimes 2 series packs are joined in parallel then shrink wrapped together. Some of my bigger jets use three 4S2p packs of 8.0 ah total capacity. The total pack is 12S consisting of 3, 4S2P

I'd call that 2 of 36S2P in parallel made up of 12S2P modules. The 36S total pack voltage is important to convey I think.

 

[JoeS]

Thank you all for your comments.

mdbuilder, yes, in the RC literature I've seen mostly first the S designation then the P; however, to me -

6S74P implies a submodule made up of six cells in series, and then 74 of these submodules in parallel.

74P6S implies a module consisting of 74 paralleled individual cells, then six of these modules in series.

In both cases, the voltage is the same and determined by the 6S designation, despite the radically different hookup.

Since we know that Tesla wires 74 (I thought it was 77?) cells in parallel (each individually fused), perhaps it should thus be 74P6S?

--------------------------------------

In my Sparrow case, I've been content to simply state that I have two paralleled strings of three 2P12S Modules in series.

...I think that would be a 2P12S3S2P Pack...

jray3, I'll drink to that! :ugeek:

 

[mdbuilder]

Think of it however you want but the standard is S first.

Why? Because it is vitally important to know the total pack voltage. Two reasons, first is the voltage of the I-Miev for instance @ 88S or 360 volts can KILL YOU. Second, if you are not dealing with a pretty pre-packaged commercial plug and play product you will burn you car to the ground if you set the voltage too high on the charger.

 

[PV1]

I thought Tesla modules still had a BMS board on each module. A member of TMC installed a solar battery system with these at the heart and simply reused the existing BMS. Unless the DIYer removed them.

As for top vs. bottom balancing, I don't see much of a difference, and the reasons I've heard supporting bottom balancing are excuses to skate by without a cell-level BMS. Besides, with CaniOn and top balancing, it's easy to see if one of the cells has an issue and loses some charge.

As for pack designations, I've always read it as x number of cells in series with y cells wired parallel. If there are multiple strings, then that'd be z number of strings in xSyP configuration.

 

[mdbuilder]

Bottom balancing let's you use 100% of what is in the pack and have a decent chance of reviving all the cells on charge since they all died at the same time.

Top balancing let's you charge to the exact top voltage you choose without a fire but you'd kill some cells if you try and extract every electron.

A bottom balanced pack charged to the # of series cells *4.2 volts is pretty likely to start an energetic fire with some cells at 4.0 volts and some at 4.5+

 

[PV1]

Bottom balancing let's you use 100% of what is in the pack and have a decent chance of reviving all the cells on charge since they all died at the same time.

Top balancing let's you charge to the exact top voltage you choose without a fire but you'd kill some cells if you try and extract every electron.

A bottom balanced pack charged to the # of series cells *4.2 volts is pretty likely to start an energetic fire with some cells at 4.0 volts and some at 4.5+

But then again, a proper BMS would prevent over/under-charge in both instances.

Either method is still limited by the weakest cell. Even if I had to run below 0% charge (obviously can't with production EVs, but possible with conversions), I'd still prefer a top-balanced pack. Then, you're only murdering a few cells and not the whole pack. One would then replace the few cells that were drawn too low with new cells, thereby possibly incrementally increasing the capacity of the pack.