Rivian battery overheating during charging in cool weather

[MountainBikeDude]

That is correct.

On the Ford F-150 Lightning, they fixed an air damn fault issue with an over-the-air update recently. The problems on the air dam are a less-than-optimal system for raising and lowering it.

The less-than-optimal system of course was not corrected with an over-the-air update. The parameters for when the warning light comes on was adjusted, so that people see those warnings far less frequently, if ever. There, customer not complaining, problem solved. Did it change the behavior of the air dam? No lol.

Since Rivian cannot change the actual physical cooling characteristics of the battery pack over the air, a similar fix can be expected. Instead of telling you why your dcfc is being throttled, they will remove that information and remove a lot of complaints in the process. There is precious little a manufacturer can do over the air to improve cooling while dcfc, except for possibly moving the entry and exit parameters.

Do I know for certain that is going to be the fix? No, I don't work for Rivian. But I was in the industry for 30 plus years, so I'm pretty familiar with how these things work.

It seems like more of a false positive or soft glitch than anything else, as charging in 30ish degrees F doesn't make sense for it to be thermal gating issue.

Sponsored

 

[R.I.P.]

It seems like more of a false positive or soft glitch than anything else, as charging in 30ish degrees F doesn't make sense for it to be thermal gating issue.

Well, I would refer you back to my original post. The pack is well insulated, and is able to shed very little of the heat generated into the atmosphere naturally. It must be drawn out through the cold plate. Sure, the conditioning system will run more efficiently in this case when it is colder, but that single path is still the only viable way to dissipate the massive heat created through dcfc.

Is this a false positive? Are the cells really not getting as hot as the BMS is saying? I would suggest that might be an even bigger problem than what is the most obvious problem; the design overestimated how much heat could be drawn out of the packs with the current design of the system.

 

[R.I.P.]

For those that wonder why the truck is overheating even in very cold weather, a good source of research is the design iterations of the Nissan Leaf. The engineers in that project grossly overestimated how much heat they could dissipate naturally; and that pack was designed to be air cooled; not well insulated like the one in the Rivian.

 

[Cosmacelf]

With Rivian's inaugural pack, they are using a sandwiched cooling plate to warm/cool the pack. This is of course going to have nowhere near the ability to wick away heat that Tesla's in-pack cooling has, so heat throttling is going to happen. The reason cold weather does not help this much is the fact that the center of the packs (they are stacked) are well insulated from the outside temperature, so the main way for the heat to escape remains that single sandwiched cold plate. This is a physical thermal bottle-neck that is designed into the system and cannot be helped.

Why did they not put cooling ribbons between the cylinders the way Tesla does? Cost, and manufacturing difficulty. I expect Rivian's packs to evolve and get better at thermal management, but this is a good way to get decent cooling without breaking the bank as they try to ramp up. It is similar to how Ford & GM are doing it, except Rivian is using Tesla style cylindrical cells.

For those watching the pack tech closely as I am, I will add that the sandwiched cold plate design has another characteristic that could easily cause problems down the road. Because there are vertical banks of 2170s both above and below the cold plate, both cooling and heating will potentially be quite different between the upper and lower packs. When attempting to cool the pack during DCFC, the pack under the cold plate will get the bulk of the cooling due to thermal dynamics. Keeping the _whole_ pack at an ideal temperature becomes impossible, the pack on top of the plate will always be much hotter.

Heating the pack will have the reverse problem, but since heating is not as critical as cooling, that should have less of an impact on the life of the packs. Just based on common sense, one would expect the packs sitting on top of the cooling plate to fail before the lower ones.

Time will tell, and I am certain that the design will evolve.

Great post. I too am worried about Rivian's single plate design. They are the only EV with this temperature control design (single plate for two layers of cells). The big problem I see is in the winter. The bottom layer gets cold since it is in contact with cold air, especially when driving. The top layer stays warm since it is insulated and in contact with the cabin. So, you arrive at a DCFC and you start charging. The bottom layer is too cold to accept a fast charge so you start heating it, but then the top layer gets too hot too quickly, so you end up with both layers fighting each other until they eventually reach thermal equilibrium. I think that's why people are seeing ping ponging charge curves in the winter.

BTW, this youtube channel did a deep dive on battery pack thermal regulation. I had originally thought that cell end can cooling was better than side cooling, but it turns out not to be since you have much more surface area in side cooling than you do when using end cooling. Add the extra layer of cells and it makes it doubly bad.


For me personally, this won't matter much since 99% of my charging is in my southern CA garage. I'm not exactly pushing the limits.

 

[Cosmacelf]

As for cold-plate vs tesla's cooling ribbons, even tesla has admitted that axial cooling is more effective than radial (to the extent that they are moving from radial to axial cooling in newer designs).

Nope. I think that's old information (maybe from Munro where he was speculating). They are sticking with radial. The latest tear downs of the 4680 Model Ys show radial cooling, not axial. I've queued up this Munro video where they talk about this very issue:

 

[electruck]

Nope. I think that's old information (maybe from Munro where he was speculating). They are sticking with radial. The latest tear downs of the 4680 Model Ys show radial cooling, not axial. I've queued up this Munro video where they talk about this very issue:

Thanks for that update. There is quite a bit of independent research out there to the contrary so we have to be careful drawing generic conclusions as specifics of the cell design such as whether it is tabless or not will impact heat generation and transfer through the cell. Most of the independent research was likely based on more traditional cell design but Tesla obviously found that they could get by with radial cooling in spite of the larger diameter cells. Sticking with a design they had already perfected also reduces overall risk at a time when they are making significant changes to other aspects of their battery pack.

 

[Cosmacelf]

Thanks for that update. There is quite a bit of independent research out there to the contrary so we have to be careful drawing generic conclusions as specifics of the cell design such as whether it is tabless or not will impact heat generation and transfer through the cell. Most of the independent research was likely based on more traditional cell design but Tesla obviously found that they could get by with radial cooling in spite of the larger diameter cells. Sticking with a design they had already perfected also reduces overall risk at a time when they are making significant changes to other aspects of their battery pack.

Tesla has a lot of experience with battery pack cooling design. Their first battery packs with the model S weren’t all that great. A model S could do one lap around a track for instance, before hitting thermal limits. The latest model Y packs are much improved.

The potential issue with the Rivian isn’t the axial cooling. It is that it has a shared cooling plate between two layers of battery cells. Even so, I don’t expect this will result in a longevity issue for the Rivian battery pack, since they seem to have adequate monitoring.

 

[electruck]

Tesla has a lot of experience with battery pack cooling design. Their first battery packs with the model S weren’t all that great. A model S could do one lap around a track for instance, before hitting thermal limits. The latest model Y packs are much improved.

The potential issue with the Rivian isn’t the axial cooling. It is that it has a shared cooling plate between two layers of battery cells. Even so, I don’t expect this will result in a longevity issue for the Rivian battery pack, since they seem to have adequate monitoring.

A single shared cooling plate isn't a limiting factor so long as there is adequate fluid flow through the plate to dissipate the heat.

 

[electruck]

The fact that one solution might be more or less effective than another has no bearing on whether either solution is adequate.

 

[SeaGeo]

The charge rate for the vehicle is controlled by the vehicle, so who's DCFC you connect it to should have no effect on the charging curve.

There's another possibility here, which is that Rivian's aren't preconditioning correctly when navigating to EA chargers because the nav thinks they are all 150kw, so it's not prepping the battery for 200+kw. This has been a suspicion I've had for quite awhile, and RIvian confirm that truck actually does take into account the power of the charger it is navigating to in their preconditioning algorithm.

The only instance I am aware of where this allegedly happened was with a Ford lightning, and it destroyed the vehicle. This is a massive failure of how the system is designed to work, and should really never happen.

Kyle has seen it happen on multiple vehicles, including his Rivian at around 70% SOC on the new signet chargers.

In this case, the vehicle is overheating during dcfc, so it's the opposite case of all your other examples.

It's also not actively cooling ahead of time with the full hvac system. Which *may* be a data gathering phase for them.

The fact that one solution might be more or less effective than another has no bearing on whether either solution is adequate.

Yep.

 

[TheIglu]

I've only DCFC'd a handful of times, but I've never run into a situation where I've seen my truck slow down it's charging rate and display a message of any kind.

Does everyone experience this? Maybe I just haven't noticed?

 

[MountainBikeDude]

I've only DCFC'd a handful of times, but I've never run into a situation where I've seen my truck slow down it's charging rate and display a message of any kind.

Does everyone experience this? Maybe I just haven't noticed?

In Kyle's demonstrations, it's happening after long drives/trailering down hill.

With what is now imminent delivery on my end my two hold ups are the potential overheating of the system, which seems to be evident with regen on steep grades scaling back. This is the biggest area I think the Enduro motor/system will benefit over the current quad being designed in-house.

Second is the tonneau. It'll for lack of a better word, blow, if shortly after delivery, they revise the power tonneau and reincorporate it into new models.

 

[TheIglu]

All my DCFC has been during long trips trailering.

I have noticed the regen doesn't work as long as it used to in previous software versions, which is opposite of what the update documentation says.

 

[R.I.P.]

This video documents with hard data the limitation of this sandwiched cold plate design. It documents how this design is unable to effectively pull down the heat from the "hot parts of the pack" (most likely the top of of the upper pack).

The cold parts get colder, but the hot parts remain relatively unaffected (hot). It is not addressed in the video, but this certainly is not good for pack health, this type of heat disparity throughout the pack. You can do your own math based on his data points on just how great this disparity becomes.

It is also a great side to side comparison of how Rivian's software push is fine-tuning what they are able to control, within the limitations of this design.

Sponsored