Pics/Info: Inside the battery pack

[glhs272]

Isn't it obvious that it is? This has been discussed previously in the Roadster threads. I think it's the Achilles heel of a Tesla battery pack. Most of our batteries will die that way - with most of the cells still in good shape and a few bricks in bad condition. For 2 years I've been trying to think of a good engineering solution to this but haven't come up with anything affordable or cheap enough.

Yes it is sorta obvious or at least I have been thinking about this failure mode. I was hoping I was wrong. It has been discussed that the Tesla pack is immune to single cell failures, but that only seems to be true for catastrophic single cell failures (i.e. sudden short) where the cell fuse blows. If the BMS system were capable of active balancing (i.e. charge shuttling) it might be theoretically able to cope with this type of failure by keeping the brick voltages up by drawing from other bricks. From a whole pack perspective it would appear that the pack self discharges a little quicker, but might be hardly noticeable. But if the BMS is only capable of bleeding off excess voltage from bricks through resistors and not actually capable pushing charge to individual bricks via dc to dc converters, then yes the pack is going to die this way. If it appeared that the pack was starting to die by doing this, it could be fixable if you were willing to open up the pack and trace down the bad brick/module. Either replace the module as a whole or dive even deeper and try to find the bad cell. Once you found the bad brick or group of 74 cells in parallel, it would be a challenge to figure out which cell is bad. But if you could, you could just snip the fuse and take it out of the system. Perhaps if you charged the group of 74 cells and then snipped all of the fuses and then measure voltage at each cell and find the bad one, then solder on new fuses to all the good cells. Or perhaps if you were to charge the 74 cell group and image the pack with a FLIR camera, perhaps the bad cell would be noticeable by being hotter than the rest. Then just snip the bad cell. Of course all of this would have to be done before the rest of the cells get too low in voltage and get damaged.

 

[spaceballs]

Sounds like interesting research, for how many cells was this done--do you have a thread or blog describing what you did and how it worked, etc.?

Sorry no blog or thread, maybe if I find enough time I should do a proper write up and do a talk at TMC Connect (if they want me).

 

[tom66]

The "solution" to this is large format cells. But they come with other hazards, namely being harder to cool (limiting charge/discharge current and lifespan -- possibly moreso than Tesla packs) and having a greater cost per kWh.

We know the Gigafactory is geared up for cylindrical cells, so I don't see Tesla changing this any time soon.

 

[spaceballs]

The "solution" to this is large format cells. ...

How does this mitigate the problem of having one cell slowly (or not so slowly) drag the rest down in that group/brick?

 

[tom66]

How does this mitigate the problem of having one cell slowly (or not so slowly) drag the rest down in that group/brick?

Large format cells that are tied in series. No parallel pairs. So no bricks to bring down.

As I mention, this solution creates more problems than it solves. Arguably the larger cell could be less reliable, due to the heating/cooling issues mentioned. It would also probably have to be pouch format, meaning lower energy density.

 

[hcsharp]

Large format cells that are tied in series. No parallel pairs. So no bricks to bring down.

That's not entirely correct. The large format cells are not that large that they can eliminate putting some of them in parallel. The only question is when and where they are connected in parallel. For example, some manufacturers may be using multiple long strings of cells in series. But these strings are ultimately connected in parallel when charging and discharging. This does not create any "solution" to the runaway failure problem that we've been talking about. In fact there's a good argument that large format cells make it worse depending on the capacity of the pack, number of series strings, etc.

As I mention, this solution creates more problems than it solves. Arguably the larger cell could be less reliable, due to the heating/cooling issues mentioned. It would also probably have to be pouch format, meaning lower energy density.

I agree on the thermal management issues.

 

[LittoDevil]

This battery package is a maintainace nightmare: how is Tesla supposed to repair a defective battery, if it has to destroy the glued cover to access the cells?!?
There must be another method.
I also wonder if there is some redundancy in such an high complexity "part", made up of around 10.000 cells; if no redundancy is present, just one failure in 10.000 will make the whole 85 kWh battery unusable and un-repairable: a 10$ failure which leads to a 30.000$ unrecovrable failure?!?
It is just not possible.

I wondered that myself.. Tesla Burlingame now has a battery room where they only remove and replace the modules but not the individual cells. If it's sealed like that.. they must replace the top with a new cover each time. From my knowledge, none of the batteries end up being a $30,000 unrecoverable failure as they are usually shipped back to the factory to get refurbished; the bad modules replaced.

Disconnecting the coolant loop from the top oddball module. Turns out I drained 99% of the coolant out.
Does anyone know what the heck kind of fitting this is?

I don't know the name of it but being an auto mechanic & auto body shop owner.. and being a mechanic.. those connectors are typically found on fuel lines on german vehicles, I've seen them on other vehicles as well but they are basically quick disconnects that are found on plastic fuel lines that connect to the fuel pump. They can also be used as air lines instead of for fluid like.. gas vapor transported to the intake manifold to be burned by the engine for emissions purposes.

Larry

 

[kennybobby]

Hey WizKid, where did you find the pre-charge resistors, what's the ohmic value and power rating, etc.? Are they in-line all the time or switched? Curious folks want to know...

ps. i think those fluid fittings look like Parker Hannafin SAE J2044 also, but can't seem to find a good datasheet for them.

 

[wk057]

There is one precharge resistor that is mounted to the mating plate (where the external connectors are), for heat dispensation presumably. I don't have it in front of me but if I recall correctly it is 23 ohms in a ~2.5"x2"x0.3" package, cased in aluminum. It was connected to the main control board in the rear of the pack to a connector labeled "precharge resistor." I presume it is switchable.

The main board also has a HV connector that has inputs that are connected in front of the main contactors (and also a set behind the contactors), so it may be that it could flow power through this resistor from the pack with the contactors open as well.

 

[offpist]

I expanded my "nissan Leaf battery module". Now it contains 70 cells, about 35KWH.
Why not simply break the Tesla battery pack down to the bare cells, then add 3.party BMS systems.. there are plenty of options out there.

Building with Nissan Leaf modules are really easy, almost like Lego

 

[JRP3]

Take apart 7,104 individual cells? :scared:

 

[LittoDevil]

Take apart 7,104 individual cells? :scared:

Hahahahahahahaha...

I guess building with Tesla's modules are easy too but much bigger...

Larry

 

[offpist]

Take apart 7,104 individual cells? :scared:

No, not quite what I had in mind.
Is it possible to rearrange all cells in one module so they are all in parallel?
It will of-course involve cutting some connections, and welding on some new ones.

Now its 4,2+4,2+4,2+4,2+4,2+4,2=25,2 V Right?
Put them all in parallel and it would be much easier to build a descent battery module out of it.

If each module was 4,2V,you connect 14 in series, and there you go.. A great 48V package.
Most 48V inverters and chargers can work well withing the range of 40-60V.

Or 7S2P to make 24V battery packs.

 

[scaesare]

Large format cells that are tied in series. No parallel pairs. So no bricks to bring down.

As I mention, this solution creates more problems than it solves. Arguably the larger cell could be less reliable, due to the heating/cooling issues mentioned. It would also probably have to be pouch format, meaning lower energy density.

Being in series, it also presents the same problem of limiting the entire pack to the potential of the lowest cell. Thus if a single large format cell begins to fail and degrade, the entire pack will be subject to the level of charge the lowest cell can accomodate.

 

[wk057]

Why not simply break the Tesla battery pack down to the bare cells, then add 3.party BMS systems.. there are plenty of options out there.
Building with Nissan Leaf modules are really easy, almost like Lego

That's basically what I've done.

See my very temporary test setup here: Plan: Off grid solar with a Model S battery pack at the heart - Page 9

The pack is 44.4V nominal in this setup (3.7V per cell) which works fine with the 48V inverters I've chosen. It seems when the voltage gets in the lower 40's the inverter output voltage sags a bit, but it still works.

 

[offpist]

Ah, I see.
But in that way it would be very difficult to make an BMS solution.

The way I suggest, would make that very easy.

 

[wk057]

Ah, I see.
But in that way it would be very difficult to make an BMS solution.

The way I suggest, would make that very easy.

The Tesla BMS has a connector going to each set of cells on each module, so, setting up the BMS wouldn't be too hard... just still need 16 setups. I plan to custom make a BMS that utilizes this connector.

 

[offpist]

That`s going to be a LOT of cables.
But yes, I guess it can be done that way.

 

[Leigh]

Obsevations from an Engineer

I've just completed a BMS for a commercial product and thought I would offer some observations.

Balancing.
Shunt balancing is the same method I'm using. Once a pack has been top balanced there is usually very little else to do. Total pack capacity is limited by the lowest capacity of a paralleled cell group.
There are ways to shuttle power around using switched capacitor, and or inductive (dc-dc) methods, but these are costly and add a whole lot of complexity to working out your failure mode analysis.
In fact, once initially balanced, future balancing has nothing to do with different cell capacities, but different internal cell resistances and leakage rates.

Reliability.
There would have been some noisy debates in the engineering team about this one. The tradeoffs of using higher capacity prismatic cells vs low capacity cylindrical cells is a hard problem.
eg. If one cell out of 7000 fails open circuit (or blows the protection fuse with an internal short circuit), the capacity of the entire pack is reduced by the number of seriesed cells. So for 16S system if one 5Wh cell fails, it reduces the pack by 16 x, so 80Wh.
Worst case is if one cell became leaky. This prematurely discharges a group (requiring constant re balancing) but with no hope of isolation the faulty cell.

There was a suggestion of having an ASIC per cell. This could disconnect the faulty cell, but on a 7000 cell pack this would add a lot more failure modes and significant cost that would be worse than the occasional cell failing.
The ASIC idea might have legs for a prismatic high cell capacity system, but not for cylindrical cells.

Effort is better spent on making high quality cells, rather than trying to work around them the complex electronics.

Future
If I had a crystal ball of what the future would be. I think the anode/cathode materials will improve to reduce impedance (so less heat in high charge/discharge), increase cycle & calendar life, and improve combustion safety to a point where where larger prismatic cells will be more viable. Having a prismatic with 100 times the capacity of a cylidrical makes it more economic to manufacture, maintain, and with more opportunity for electronics to intervene.

Wearing my systems engineering hat, the current Tesla solution 'feels right' given the current state of tech.
I think the genius of Musk is having both a good understanding of engineering AND the ability organize capital & labour. It is rare to find both skills in one person.

 

[JRP3]

Data point regarding cell capacity from the AAB report on Tesla:

Cell Capacity, Ah, Rated 3.25 Actual 3.1 Battery Capacity, kWh, Rated 85 Actual 80

http://www.advancedautobat.com/indu...ort/Extract-from-the-Tesla-battery-report.pdf