One Factor that doesn't Limit my appreciation of this channel is Jordan's transparent willingness to admit what predictions he got wrong and why. In a world of influence and fakery, this is both factually invaluable and appreciated. Astonishing to me that the Channel doesn't have a million subscribers.
@@thelimitingfactor Musk would approve that you are pushing the limits, taking risks. Science is about being less wrong, about useful predictions, understanding. Full credit to Tesla engineering team for being totally amazing.
Jordan’s intellectual humility personifies the difference between scientists and Wall Street analysts - who are never wrong about any thing at any time …
Thanks once again for your insights. Surface area of contact with the thermal sink is the key determinant of sink efficiency. That makes sense. And the tabless design reduces resistance and thereby reduces point thermal spikes, thermal heterogeneity, and accelerated electrode failure.
Actually NO! Your last sentence is accurate but your take on 'Surface area' efficiency is 'out of context' in this case. When 'all other things are constant' (heat sinks being equidistant from the source etc) then contact surface area makes a difference. However, the Top & Bottom plates are much closer to the heat source (current collector inside the can) than '1 rectangular patch on 1 side'. The video's take that Top-Bottom cooling somehow 'orphans' the sides is ridiculous. Most heat will flow from current collectors to the 'infinite-tabs' connected to Top-Bottom caps....leaving very little to flow thru the dozens of insulating layers leading upto the side walls. Let's assume that heat from current collectors somehow permeates to the sides (in large quantity), then if side walls can wick heat from the Top-Bottom to provide efficient (?) cooling (as described in the video), then it's only logical to expect the reverse flow (that Top-Bottom will also be able to wick heat from the side walls). Net-net, Top-Bottom cooling of 4680 will always be more efficient than (one) side cooling. However, the choice made by Tesla might have been driven by other considerations (like safety....ability to vent a thermal runaway via the bottom etc). Even though Top-Bottom cooling maybe the MOST efficient, the side cooling (via ribbons) may have achieved REQUISITE cooling while providing design flexibility to take care of other aspects (safety against thermal runaways etc). Note that Top+Bottom cooling means heat is wicked away via all metallic contacts. Side cooling ( or for that matter Top-only or Bottom-only) essentially means that one set of current collectors are directly cooled, while another set of current collectors have their heat wicked thru the insulation material to the adjacent current collector of opposite polarity). (Also to be noted is that Top-Bottom cooling that facilitates a more pristine honeycomb packing will also be structurally MORE strong. But again, Tesla might have attained REQUISITE strength in the current design that they felt comfortable to heed to other aspects viz safety)
Great video. I was also under the impression that plate cooling for cylindricals was the best approach. Also worth mentioning that teslas 4680's uses ~600 micron cell cans, which would greatly increase the thermal conductivity of the can itself
Can you help me understand this better? Are you saying that something about the outer cell wall thickness makes side cooling better/worse? As I understand, the "structural" 4680 can is thicker than traditional cells. Doesn't thicker wall slow down heat transfer? Or are you saying the cell side wall thickness much thinner relative to end cap thickness? Just trying to understand how wall thickness plays into the system.
@@oof_Dad good question. What i meant was that the added thickness (for structural reasons) (compared to other cylindrical cells) helps with thermal transfer as you've got a larger number of atoms which can transfer the random kinetic energy throughout the entire cell can body, despite having the source of the heating/cooling being further away. Usually, adding material would increase the thermal mass and slow down the temperature change. The kinetic energy transfer would still be the same, but the temperature change per unit kinetic energy gain/loss would be lower. The less atoms you've got, the more kinetic energy per atom, and thus the higher the temperature gain/loss per unit kinetic energy added. Energy stays same. However, the added thermal mass in this case is negligable, meaning that the only appreciable effect would be that of a greater thermal transfer capabillity. By having thicker sidewalls (as a result of a thickner can), you increase the thermal transfer to the entire cell body.
At @15:15 you state that the heat dropped by about 60%, which is correct based on the math shown. But at @15:30 you multiply by 0.6, which would be only a 40% reduction. You should have multiplied by 0.4, or divided by 1.6. The conclusion you reached was that heat losses come down from 5% to 3%. But the math would bear it out that you go from 5% to 2%. While that's a 1% difference, the difference between 2 and 3 is 50%. Great video so far, loving it always.
Along with the structural benefits of ribbon cooling and the fire mitigation benefits, it is just a much more feasible manufacturing task than assembling hundreds or thousands of cylindrical cells in a bottom cooled pack. Especially given the need for standard separations between cells and placement for the collector connections, it is just much easier to glue the cells to a ribbon cooling tube and make bandoliers then to place hundreds of loose cells into a pack precisely and try to assemble the pack from there without jostling them out of place or tipping them.
It’s possible there is also another couple of benefits to ribbon cooling. The ribbons can act like a stiffener to the battery pack, providing internal support. This in turn reduces amplitude of vibrations, which in turn reduces forces acting on the battery connection. Ultimately providing better longevity of the pack. Where does this assumption come from? My first Job after school was as a sheet metal worker. Square ducting is often folded diagonally to increase stiffness and reduce noise. Noise is vibration. A straight ribbon would be floppy, a formed ribbon will be stiffer.
I recall Sandy Monroe stating, about a year or two ago, when the 4680 was first being mooted that he felt the power dissipation aspect of the 4680 was the most compelling part of the design, even before any particular chemistry. And this video supports that view.
Something nobody seems to ever mention is the relative simplicity of the manufacturing equipment required to make large cylindrical cells compared to pouch types. The simpler the equipment, the more reliable and the cheaper it is to produce. It's also easier to ramp up the production rate than all of the stop-start motions and complex alignment moves required for pouch cells. I really can't see Tesla changing from a simple to a complex line which much surely take more factory space, even if there are marginal gains in other areas. Elon keeps saying that the product is the factory, and the large cylindrical cell format fits that narrative perfectly.
Excellent stuff ! I had to specify and manage the battery "fleet" at a robot company. So batteries are eternally fascinating (to me !) . One thing I suspected, and AI2 sort of proved, is that Tesla's ability to model the real world is unrivalled ( nobody really mentions this rather important moat). That is a mix of software, mathematics, physics and engineering expertise, to model whatever they need to optimise - rather than rely on "expert opinion". What's interesting here is the "bias towards the proven result" - once you realise the reasons why side cooling is better - "Well it's obvious! "...(except it wasn't at the time...)
And they said "it's not that big a deal!" What do I feel like every other manufacturer is basically going to be like "make the cell bigger" and then panic when their fleet is on fire because the customer floored it or put it under real world load. Ladies and gentlemen DUMP your money into Tesla. At least as long as Elon stays.
@@thelimitingfactor "The first generation Nissan Leaf was air cooled"... All generations were convection cooled. Only the Nissan e-NV100 was fan air cooled.
With base plate cooling, the cans are attached to the cooling plate rather than a structural plate. I imagine the latter yields better structural strength/rigidity.
Well done. Thanks Gordon. Regarding Thermal function of the top/bottom Plates. My take is because your Tested 4680 Cell was not used under pressure, in a car, prevents us from seeing the Actual affect of heating cycles. Hope that @Munrolive will dissect one.
Imo, this all ties in with the "odd" charging curve displayed by the "Korean cousins". They seem to have a "front half,/ rear half" temperature differential which they can't control.
Here’s where Tesla’s “Fail Fast” philosophy makes a difference. Tesla’s engineers very likely just came up with several different ways to cool the cells, including ribbon cooling, and just measured the temp difference under load. They didn’t write a paper, they just found out what worked best via direct experimentation. It is often the case that you can make something and tear it apart and remake a new design of it five times in the amount of time you could verify something theoretically. In other words, building something and live testing it in the real world is often faster than theoretically modeling something before building it. Also, building things live gives opportunity to stumble upon new findings and knowledge that might well have been missed by modeling something to get a result before building it. I worry a lot about todays engineers getting too far away from a hands on understanding of materials. It is better to have an idea and build it and test in the real world and refine and rebuild than it is to computer model something to death before ever building it. Sure there are exceptions such as when you need to know if something is possible at all, but hands on is the quickest and best teacher.
Surely this heat management advantage is going to play into improved charging characteristics. Either for one of charging or constant cycling such as the Tesla-Semi is likely to experience.
I definitely think that the thicker casing of the structural 4680 cells will help spread the cooling more evenly around the entire circumference of each cell, helping to negate the increased ratio of diameter to length found in the 4680 (as well as the structural benefits).
This is amazing analysis. Rivian and Lucid used plate cooling on their EVs. I think that they will have performance issues when performing back to back drag races. Tesla Plaid does not have this issue since they fixed the serpentine ribbon cooling to a parallel ribbon cooling like the Model 3. The optimal design for heat dissipation then would be the tabbed 18650 cell with parallel ribbon cooling based on your analysis. Great job Jordan.
It is interesting that Lucid seems to not have overheating issues despite their bottom cooling design. They started out designing battery packs for EV racing so they must have some other innovation that results in better cooling.
Interesting. Note that Peter Rawlinson threw side cooling under the bus in favour of Lucid's side cooling. His argument was that it is difficult to get reliable contact with the curved surface area of the battery. I was not convinced.
@@bru512 Peter is right, it is hard to get reliable contact. Tesla bought the company that was making the cooling tubes for them because quality control was not up to the required level. Basically there is a coating on the cooling tube that is cooked on, and it has to be cooked just right. Too "raw" and the surface is slippery and the glue that bonds the cell to the tube slips off. Too "well done" and the coating cracks and peels off the tube. Then the glue used has to be the right mix of the two ingredients, the applicator tube has to be properly maintained and changed often as it will clog over time, etc. One guy who was what they call a CTA (cell tube attach) worker blew off changing the application tips his whole shift, they had to bring finished battery packs back from Fremont and he got fired. Yeah, just like the "experts" said, the Model 3 battery packs being made on an automated line was impossible, and when they say that just add "for anyone except Tesla" and you may have a true statement.
@@bru512 well, the cars are in the wild now and I don't think there have been reports of acceleration being limited due to heating. I may have missed them, of course. One thing that may be helping Lucid avoid heating issues is that the pack is enormous at 118 kWh, so the acceleration heats each battery cell less because there are more of them.
Tesla's battery cells don't really need maximum thermal performance typically. Cells don't get that hot normally unless in extreme conditions like at the track perhaps. Also, uniformity is good when cooling or heating, you don't want the bottom of the cell to be different than the top half, so ribbon cooling cools the entire jacket which then cools the entire cell, making it uniform temperature. Good thermal management is needed for long term performance of the cells, but having maximum thermal conductivity is not even really necessary.
I am probably a bit late to mention that if doing base cooling, and there is a thermal runaway event, the expansion has to go down through the base, destroying the cooling at a time when it is needed most
Great overview! I hope someone can further clarify the "peak voltage power delivery" aspect of the 4680 vs 2170. Currently a Model 3 loses massive amounts of horsepower as battery SOC gets lower. I assume it's due to rapid voltage sag. Range may be unaffected but a Performance Model 3 with 30% SOC drives nothing like one with 85%. Will the 4680, by it's design, be more able to perform similar to the Tesla Plaid, where the car performs steadily until a much lower SOC is reached?
Power throttling is a factor of inverter design. Power = Voltage × Current this means you can maintain power even with dropping voltage by increasing current draw and the limiting factor for current draw is the inverter. The tabless design will increase max current with in the cell but I don't think Tesla is drawing max current from the cell.
You cannot see which exact cell is defective from a BMS perspective, you can only see that for each voltage set of cells. A faulty cell will drag down voltage of the whole set as they are wired in parallel. It will also limit how much that set can charge. To determine which specific cell in a set is defective, you have to disconnect the cells and test them individually. Say your battery has 4,416 total cells arranged as 96s46p. If any one of those 96 serial sets of 46 cells in parallel is significantly different than the rest of the battery pack, the entire serial string capacity suffers. This is a disadvantage of small cells (e.g. 18650), as there are so many more of them increasing the likelihood that one will fail compared to 4680s. If an individual cell completely fails with a short - that is actually a better case. The fuse link will blow and you are just out 1/46 of the capacity of the battery as it is taken out of the entire circuit. Same thing if it fails with an open, although in that case no fuse will blow.
@@thelimitingfactor Yes, one cell can dramatically impact pack level performance. But you cannot electrically measure the difference between that cell and the rest of the cells in the same voltage set with a BMS. What you can see with the BMS voltage probes is a difference between that set voltage and all the others, as you showed in the video.
Great video man. I wonder if this will be the long term or shorter term cooling solution. I would imagine it would continue to evolve as new silicon doping and material changes with v2, v3, etc if they are indeed serious about million mile battery. Also big congrats on closing in on 100k subs!!
With higher amount of cycles for LFP + lower degradation due to better thermal management of the tables design these 4680 batteries will likely outlive other components of the car enabling reuse of battery packs in other cars and/or in future storage packs. While simultaneously giving more power as the shorter travel path of electrons enables higher current flow increasing the acceleration possibilities and also the charging speed possibilities. Looking forward to cars being charged at vastly shorter times without degrading them. Will likely be even more important for Semi & cybertrucks than current passenger models. Thank you for your very detailed in depth videos! This is braingasm for a chemical engineer 🤩🤓
You'd think! But as Drew Baglino has said, the batteries have already had issues with the cells outliving the pack. Battery packs are a difficult challenge from a durability point of view as well.
So now Im thinking that the extreme thickness of the material for the 4680 cans is not just for structural rigidity, but also to increase the thermal conductivity of the cans themselves.
I wonder if there's a way to use the natural void in the center of the cell for liquid cooling? This would allow the cells to be closer together creating a higher with volumetric energy density pack. Maybe you could even have a passive mechanical thermostat in the center of each cell, allowing each cell's temperature to be controlled as an individual.
I thought I heard ( but I am not sure from where) that with generation 2 batteries and the new 3 stages combined rolling equipment they were going to omit the copper current collector and laser weld the copper foil turnovers directly to the can base.
I would also have thought Tesla would switch to "bottom/top cooling" since it's far less complex to assemble. The only thing I had my doubts about was the height, because the cells are now even longer, which would make the floor structure difficult for a sedan. Nice work!
A battery is not 95% efficient, the graph you are showing to this is the on board charger efficiency A battery 21700 has 15mR milliOhm of resistance, with a current draw of 0.2-C in high-way driving conditions, this is 4500mAh*0.2 = 0.9 A per cell = 12mW loss for 5h = 60mWh lost, on 16.2 Wh total cell (3.6V, 4.5Ah) = 0.37% lost = 99.63% efficiency of battery cell only! Battery is roughly 99.1-99.5% efficient if you include bus bar and cabling losses in battery pack assembly
When I spoke to battery scientists about it, I usually was advised of an efficiency at the battery level of 90-95%. The charging equipment was also showing in the image (also 95%).
@@thelimitingfactor that is true newer battery chemistries since 2018 model 3, now have power NCA and NMC cell chemistries where the battery is designed more to efficiency, this leads to >97-98% battery efficiency at least in regular power levels. during supercharing this drops way in of course, due to the large currents and thus resistances that add up. OBC charger is sometimes 93% some products 96% while the SiC inverter of model 3 achieves ~98% efficiency, and newer versions even 99%
A very illuminating explanation... confirming also that Tesla's engineers considered all possible functions (thermal, electric, structural, ...) of every part in order to improove vehicle's efficiency...👍🤙
The way that the current collector jumps from the outer edge to the core, skipping the middle layers, is exactly what you would want to reduce the temp increase in the center. If you contacted all the layers then the outer layers would "use up" the cooling potential, make it less cool where it hits the center hotspot. There may be an analogous electrical advantage if the electrical resistance to the center is also higher, though not obvious why that would be so in the tabless design.
Sometimes the initial choice for a particular design maybe the best over time, even for a company which keeps refining its design. The side cooling ribbon design is still bringing the best result for Tesla. However Lucid is doing its cooling through the bottom end of it cells which does not make too much sense in terms of surface contact area!
Maybe this is a dumb question, but why packing thausen of cylindrical batteries vertically when you can stretch the battery to 2 meters long pipe like design, and packaging in a horizontal styles, like a floor made of pipes?
Also, cooling between cells maximizes cooling while minimizing cooling lanes required. having coolant flow on the bottom or top of the pack would only cool the cells on one side, and need to be insulated on the other side. (you don`t wanna cool/heat the structure around the cells) if the cooling channels are not the full height of the pack, they probably won`t hurt rigidity, because the lever is the highest on the top and bottom of the cells anyway. think of hollow structures or half-timbered structures.
Hmm, could they ever do center cooling sense the middle of the can is hollow anyway? Im having trouble putting the idea together in my head but im sure someone has thought of it.
Greta video but I would like to point out that plate cooling also utilizes the Can of the cell to cool the sides of the cell as well, Plate cooling allows the copper and aluminum conductors a less Thermally restricted path for heat transfer whereas in just side cooling 20 percent area the heat has to travel through multi-layers within the roll of the cell to cool. This is still effective but the difference is seen when the cells are stressed thermally in high-current modes of operation. This is why Ribbon cooling is not as effective in race conditions or long-duration high current draw situations. Of course variables such as coolant temp can factor into this but as a researcher who has experimented directly with cylindrical cells , Plate cooling a cell that is designed for it is more effective at heat transfer. I am curious as to why you did not use the Lucid module design ( a plate-cooled cylindrical cell ) in any of your comparisons as that would have been a more comparable difference between the 2 types of cooling methods.
Thanks a lot for the video and for sharing the reference. You can still use the explanation showed here in order to demonstrate that the prismatic can be as good as a cylindrical 🙂. The conductivity alone is pointless, and using the thermal resistance is more realistic for choosing the best cooling direction ( Thickness/(Conductivity*Cooling_surface). In the case of the cylindrical cell, the radical conductivity is quite poor compared to the height direction. But the radial direction has higher surface. There is a possibility to make the top/bottom cooling better but due to cell integration and important contact resistant in the height (gasket, gaps, electrical insulator) , the height is less interesting. Finally in the case of the LFP from CATL , you can apply the same reasonning and due to the use of the LPF you can have a higher Tmax overall and you do not need higher cooling surface. The sum of improvement of TESLA is driven by cost per performance and tooling is one of them. My guess is that there is less ivestment by staying in cylindrical instead of switching to another form factor. The increase of cell size has been done by CATL and BYD already in 2018 and they have also updated tabs and the number of JR inside the cell in order to cope with the thermal drawbacks
Sure thing! Depends on what your goal is. A full form factor comparison can't be done the comments here. I already point out an upcoming video that Qilin has better thermals than 4680 structural.
I believe the Tesla battery pack design consist of a group of cells connected in parallel and then these groups are connected in series. Within a paralleled group of cells the BMS doesn’t know anything about the individual cell, just about the groups. I’m not certain if this affects any of your conclusions in the video, though. 3:30
🤗THANKS JORDAN AND YOUR PATRONS 👍 GREAT EXPLANATIONS ,EASILY UNDERSTOOD BY LAYPEOPLE…🤔 It would be great to buy you lunch someday,since I am in N.E. OHIO…IF YOU COULD SPARE THE TIME 🤗💚💚💚
n March 2021, Optodot was issued U.S. Patent No. 10,950,837, titled “Methods of Producing Batteries Using Anode Metal Depositions Directly on Nanoporous Separators.” Third Generation NPORE® ECS Electrode Coated Separator technology, developed with funding by the DOE, aims to reduce the cost of manufacturing lithium-ion batteries and the inactive components cost by 20-40%, while improving battery safety, lifetime, and energy and power density. NPORE® ECS incorporates new inactive components of separator, current collectors, and termination materials, and utilizes a simpler and faster battery assembly process. With ECS, electrodes-for example, lithium metal anodes-are directly deposited onto the separator to form a separator/electrode stack.
Does this mean the cells in the centre of the Pack will degrade quicker than the ones on the outside of the pack? or are the cooling channels routed differently within the ribbon to compensate for this
US Patent No. 11,387,521, issued July 12, 2022, is directed to an advanced nanoporous separator for lithium-ion batteries which has far superior safety properties such as better heat conductivity and non-shrinkage compared to conventional separators currently used in lithium-ion batteries. US Patent No. 11,387,523, issued July 12, 2022, is directed to a new generation lithium-ion battery in which the cathode is directly coated onto the separator. This technology reduces the complexity and expense of the equipment used to fabricate the battery, for flat and prismatic batteries that can be used, for example, in electric vehicles. This US patent also enables alternative manufacturing processes, such as where a non-dried, wet cathode is applied to the separator.
I would have also gone for a top/bottom cooling. In a uni e-motorcycle competition i designed the battery pack cooling i such a way, not appreciating the full jellyroll + can approach since the infor the papers showed the later years was one approaching the jellyroll alone, with very poor radial conductivity. Also, we weren't focused on durability buy max T in 5-10 battery cycles. For the new prismatic cells bottom heat sink utilization was a no brainer
Cylindrical 2170 cells have 10x better cooling axially compared to radially. Tabless improves that significantly. Plate cooling has a similar thermal performance as ribbon cooling up to the can but ribbon cooling performs better at spreading that heat around both bottom and top of the cell better.(area is less for plate cooling but epoxy gap is smaller and more consistent so overall thermal interface is about the same. Tesla isn't able to get the ribbons to form exactly to the shape of the 21700 cell. 4680 is a huge improvement on many fronts. :)
Incredibly informative and easy to understand. I'm not a chemistry geek, so there are times when the explanation flew over my head. But overall, it confirms my faith in the 4680 battery, especially if the Tesla owner plans on keeping the car for a long time
In summary, "Ribbon cooling provides greater surface area, which results in higher efficiency of cooling". However, increasing cell size 5x also increases thermal mass. Meaning the center of the cell is cooled exponentially less than the outer layers of the cell. In theory, tabless technology coupled with ribbon cooling applied to smaller cells like 2170 & 18650 would be the most efficient way to maintain thermal management of the cell. What this means to me is that thermal cooling for the cell/pack is not Tesla's priority. The more important variables to consider, which Elon emphasizes consistently, is cost and scalability. I think 4680 would not have been possible without the tabless innovation and its impact on thermal management. Tabless cells will prove to be the single greatest innovation in batteries due to its low resistivity = less heat generated and the increased rate of cell fabrication. The next innovation will be the solution to high energy dense cells using silicon. Exciting times we live in!
Awesome breakdown, thank you. Here in southern Appalachia we heat exchange with our stills the same way, copper tubing with max thermal exchange through the walls of the tube, not the ends. ... . .
None of the Nissan Leaf's, including the second generation, have active thermal management. This is their Achilles heal, as it causes their packs to last significantly fewer miles than other EV's, and Nissan is obtuse regarding this subject.
Second time watching through. Takes awhile to get all of it and I'm sure I'm not there yet. BTW, could we please have some color choices for the hats. Please.
Not sure why they stopped at 46. They could have gone for 8080 with a hollow core instead. This would give them cooling lines through the whole core and 80% of the circumference. It would also make it trivial to connect those in series, as they line up that way automatically when threaded onto the central cooling pipe.
Can you do an analysis on StoreDot and its battery chemistry? Can you confirm(debunk) the validity & effectiveness of Silicon based anode NMC cathode? Is it a sham or is solid state truly be the future of battery innovation?
If it can conduct electricity, it can conduct heat. The current collectors at the top are wicking away heat then conducting over to side can. Dual purpose for sure. The tabless design is also drawing heat out from the center for same reason ohmic resistance is lowered. Did not surprise me that side cooling works.
Do you know of any studies showing which type of batteries are lighting up and what the conditions were? We've seen plenty of fires in line but nothing about which ones, if it was design problems or external. Thanks for your work.
Question: the cells are encased in some tough polymer as the Munro tear down showed. I get that the polymer has limited cooling capacity, unless Tesla designed that material to absorb some heat. Once warmed maybe any potential cooling bennies are negligible?
Nice job! I think you may have overestimated the thermal conductivity along a thin sheet of electrode and underestimated the thermal conductivity through a thin plastic sheet. It would be interesting to see if the people doin the modelling published numbers for that.
Thermal management is all for fast charging. I don’t plan to fast charge my Tesla almost at all. I could wait a longer time I fast charge or could fast charge just the minimal amount.
I built a LFP battery for a RV that gets used daily - 6yrs old now - new cells were Bottom balanced @2.75 volt - cell balancer units were not reliable at the time so did not install one - battery discharge to 40% daily - solar charge - I check battery cell balance every few weeks - cells hover around 4mv difference - Amazingly Good
Ribbon cooling is a little bit cheaper to implement in manufacturing. I'm pretty sure that's the only reason because the electrode itself is actually a long flat ribbon that's been coiled up which means that cooling the side is not actually cooling the side, you're cooling the ends essentially. The heat actually can travel faster along the width of the ribbon up to the top of the cylinder and then out to the sides because of the larger thermal mass then it does around the coil a bunch of times to the outside. Cooling the outside of the cylinder is easier than cooling the top and the bottom together and cooling the side cools it roughly equally. It's pretty good, it's pretty cheap, it's pretty easy to do, pretty reliable, and overall just a good balance of lots of factors. It's not really the best in any regard but without spending ridiculous money this is kind of the best compromise for the money
Given that single cell failures have a disproportionately large whole of vehicle cost, why is per-cell monitoring and isolation not built in? I can imagine that a cell absence would affect battery bank voltage and current characteristics, but are there not simple electrical solutions to those?
Excellent analysis. What you did not say is that one end of the battery is great for cooling, but at the other end there is the electrical connections for both positive and negative. Perhaps also some thin wires which monitor the performance of each cell. So, it will both add complexity and be of less value thermally, perhaps not worth the effort, to extract heat from both ends of the battery in the end cooling scenario.
Eh, I'm surprised why engineers haven't just enclosed top and bottom of cells to make it water proof, pack batteries closer so body of the cell creates channels, directly cooling cells and in case of thermal runaway of single cell you have A LOT water right next to cell to boil off before endangering nearby cells
also, the bottom of the cell has a noticeable airgap between the active materials and the casing, thus really not effective cooling area Vs side. Cooling from the top gets challenging as that's where the electrical connections are made for positive and negative welds.
Bigger batteries with better cooling mean that you can potentially charge every cell on its own, and potentially less cells can be bunched into a group/pack, the ideal being each cell a pack.
There often discrepancies between theory and practice. Only by testing can one know for sure and it would appear that Tesla has left nothing to chance and chosen the optimal cooling strategy.
The model 3 LFP pack is already using large format prismatic cells, Munro showed one in one of it videos of him visiting a vendor. I couldn't find any other sources to confirm it, nor did they go into any significant detail like how it was cooled iirc.
Let me add to the rationale in this video. (1) The anode foil wick heat more effectively towards the base plate only to be impede by the electric conducting wafer (a heat insulator) leaving a high thermal gradient between anode foil and base plate. Remedy? Solder all anode flaps at the base plate which will not set back electric current. (2) given no soldering option, heat transfer must use radial path towards cylinder wall. Heat transfer radially through chemical and insulation is undesirable but it benefits from a higher wall area as opposed to base plate with a wafer heat insulation. With better coolant circulation we can pull it off, heat in the core. (3) per base plate cooling architecture - only a fraction number of cells (cylinder wall and plate) can be directly connected to the chassis to benefit a direct heat path to a metal structural plane with coolant channel under breath. The rest are (electrically insulated) that impedes heat flow out of the base plate. Heat at the base plate must migrate around to cylinder wall to be wicked off. Hence the heat sink via cylinder wall become a prevalent architecture.
Lucid battery pack video stated that it is difficult to make the ribbon channel to be in full contact with the cylindrical cells in mass production. Instead of 20% of the area as you stated, you may get 1/4 of 1/3 of it if the ribbon is misaligned with the cell cylinder. How did Tesla solve this issue? Maybe worth to do a video on that.
Why arent cells built and cooled internally. Imagine a copper or alumium pipe of diameter 6mm with 1mm walls (or whatever size is appropriate) and the active battery materal is wound around this pipe. You could build extrmely long bayteries. Rather than 46-80 maybe even 50-800 or even bigger And then you just run your cooling internally
I believe your original assumption concerning end verses side cooling to be correct when it comes to end/tab cooling, though side cooling could potentially be superior if implemented correctly. The problem as I see it is that Tesla’s ribbon cooling solution cannot achieve equal cooling to all the cells, meaning the cells at the intake side will experience lower coolant temperatures compared to those cells at the outtake side of the pack. The other problem is that the ribbon as you have explained is only contacting at best 20% of the cells side area, a further problem is how to guarantee adequate thermal contact between the cooling ribbon and the cell side cylinder wall. No matter what, you have given me something to think further about.
I've looked into that, very little temperature variance between the intake/outtake is a small fraction of a degree. Also, that was perfected even more with the intra-ribbon loops.
If a pack has faulty cells, can Tesla disable and isolate part of the pack? Could this explain the weight of the Austin Model Y 4680 structural battery pack as per Cleanerwatt's analysis?
Great cooling video: Question: Why have they not run the cooling in a more serpentine fashion to create a greater surface area for cooling?? ~~~~~~~~.. One cooling arm could cool two row of batteries with more than 50% of the can surface area in contact. I realize that it would add a small amount of weight, but this seems negligible compared to the battery life savings with improved thermal management.😀 Thanks
That's a million dollar question! We don't know their current baselines. I'm assuming they made the ribbon decision design on balance and it could be an area of future improvement.
Ty for taking the time to so clearly and professionally lay out this presentation. The graphics and production quality hit way above what you'd expect given your subscription count. I know this took a long time e to put together. I learned a lot. I hope your channel grows. Subbed.
One Factor that doesn't Limit my appreciation of this channel is Jordan's transparent willingness to admit what predictions he got wrong and why. In a world of influence and fakery, this is both factually invaluable and appreciated. Astonishing to me that the Channel doesn't have a million subscribers.
It makes the journey a bit more interesting when you screw up, lol. Not that I plan on making a habit of it.
@@thelimitingfactor
Wouldn't exactly call it a "screw up"!
@@thelimitingfactor Musk would approve that you are pushing the limits, taking risks. Science is about being less wrong, about useful predictions, understanding. Full credit to Tesla engineering team for being totally amazing.
@@thelimitingfactor Seriously nice information, I'm gonna become a patron. This is gold.
Amazing to see Tesla chose the correct form of cooling since the beginning in 2012. The power of first principle thinking.
Kudos to you for being transparent about being wrong in the past, and correcting it.
100%
Totally, this gives Jordan huge street cred!
Keep up the great work. If your understanding changes/improves we all win!
You mean, not anticipating that Tesla would go with a legacy technology ;)
Jordan’s intellectual humility personifies the difference between scientists and Wall Street analysts - who are never wrong about any thing at any time …
Thanks once again for your insights. Surface area of contact with the thermal sink is the key determinant of sink efficiency. That makes sense. And the tabless design reduces resistance and thereby reduces point thermal spikes, thermal heterogeneity, and accelerated electrode failure.
Great summary!
Actually NO! Your last sentence is accurate but your take on 'Surface area' efficiency is 'out of context' in this case.
When 'all other things are constant' (heat sinks being equidistant from the source etc) then contact surface area makes a difference. However, the Top & Bottom plates are much closer to the heat source (current collector inside the can) than '1 rectangular patch on 1 side'.
The video's take that Top-Bottom cooling somehow 'orphans' the sides is ridiculous. Most heat will flow from current collectors to the 'infinite-tabs' connected to Top-Bottom caps....leaving very little to flow thru the dozens of insulating layers leading upto the side walls.
Let's assume that heat from current collectors somehow permeates to the sides (in large quantity), then if side walls can wick heat from the Top-Bottom to provide efficient (?) cooling (as described in the video), then it's only logical to expect the reverse flow (that Top-Bottom will also be able to wick heat from the side walls).
Net-net, Top-Bottom cooling of 4680 will always be more efficient than (one) side cooling. However, the choice made by Tesla might have been driven by other considerations (like safety....ability to vent a thermal runaway via the bottom etc). Even though Top-Bottom cooling maybe the MOST efficient, the side cooling (via ribbons) may have achieved REQUISITE cooling while providing design flexibility to take care of other aspects (safety against thermal runaways etc).
Note that Top+Bottom cooling means heat is wicked away via all metallic contacts. Side cooling ( or for that matter Top-only or Bottom-only) essentially means that one set of current collectors are directly cooled, while another set of current collectors have their heat wicked thru the insulation material to the adjacent current collector of opposite polarity).
(Also to be noted is that Top-Bottom cooling that facilitates a more pristine honeycomb packing will also be structurally MORE strong. But again, Tesla might have attained REQUISITE strength in the current design that they felt comfortable to heed to other aspects viz safety)
Great video. I was also under the impression that plate cooling for cylindricals was the best approach. Also worth mentioning that teslas 4680's uses ~600 micron cell cans, which would greatly increase the thermal conductivity of the can itself
Can you help me understand this better? Are you saying that something about the outer cell wall thickness makes side cooling better/worse? As I understand, the "structural" 4680 can is thicker than traditional cells. Doesn't thicker wall slow down heat transfer? Or are you saying the cell side wall thickness much thinner relative to end cap thickness? Just trying to understand how wall thickness plays into the system.
@@oof_Dad good question.
What i meant was that the added thickness (for structural reasons) (compared to other cylindrical cells) helps with thermal transfer as you've got a larger number of atoms which can transfer the random kinetic energy throughout the entire cell can body, despite having the source of the heating/cooling being further away.
Usually, adding material would increase the thermal mass and slow down the temperature change. The kinetic energy transfer would still be the same, but the temperature change per unit kinetic energy gain/loss would be lower. The less atoms you've got, the more kinetic energy per atom, and thus the higher the temperature gain/loss per unit kinetic energy added. Energy stays same.
However, the added thermal mass in this case is negligable, meaning that the only appreciable effect would be that of a greater thermal transfer capabillity. By having thicker sidewalls (as a result of a thickner can), you increase the thermal transfer to the entire cell body.
At @15:15 you state that the heat dropped by about 60%, which is correct based on the math shown. But at @15:30 you multiply by 0.6, which would be only a 40% reduction. You should have multiplied by 0.4, or divided by 1.6. The conclusion you reached was that heat losses come down from 5% to 3%. But the math would bear it out that you go from 5% to 2%. While that's a 1% difference, the difference between 2 and 3 is 50%.
Great video so far, loving it always.
Along with the structural benefits of ribbon cooling and the fire mitigation benefits, it is just a much more feasible manufacturing task than assembling hundreds or thousands of cylindrical cells in a bottom cooled pack. Especially given the need for standard separations between cells and placement for the collector connections, it is just much easier to glue the cells to a ribbon cooling tube and make bandoliers then to place hundreds of loose cells into a pack precisely and try to assemble the pack from there without jostling them out of place or tipping them.
Jordan your knocking out videos with the quickness! Love your stuff! Keep up the great work!
It’s possible there is also another couple of benefits to ribbon cooling. The ribbons can act like a stiffener to the battery pack, providing internal support. This in turn reduces amplitude of vibrations, which in turn reduces forces acting on the battery connection. Ultimately providing better longevity of the pack.
Where does this assumption come from?
My first Job after school was as a sheet metal worker. Square ducting is often folded diagonally to increase stiffness and reduce noise. Noise is vibration.
A straight ribbon would be floppy, a formed ribbon will be stiffer.
Bingo! Thanks for your thoughts
I recall Sandy Monroe stating, about a year or two ago, when the 4680 was first being mooted that he felt the power dissipation aspect of the 4680 was the most compelling part of the design, even before any particular chemistry. And this video supports that view.
Thank you!
Sure thing Paul!
Great and educational video. Thank you Jordan for your hard work and dedication.
There is something soothing about your voice that it help me fall asleep in the evening 😅
😄 Quite common.
Very interesting. I was under the impression that plate cooling was superior due to those plastic separators so the nuance is much appreciated.
Something nobody seems to ever mention is the relative simplicity of the manufacturing equipment required to make large cylindrical cells compared to pouch types. The simpler the equipment, the more reliable and the cheaper it is to produce. It's also easier to ramp up the production rate than all of the stop-start motions and complex alignment moves required for pouch cells. I really can't see Tesla changing from a simple to a complex line which much surely take more factory space, even if there are marginal gains in other areas. Elon keeps saying that the product is the factory, and the large cylindrical cell format fits that narrative perfectly.
I've mentioned it in the past. But this video isn't about the cell manfucturing cost.
It's the basic principle.
Large amount of GWh (TWh) from the smallest possible factory.
Excellent stuff ! I had to specify and manage the battery "fleet" at a robot company. So batteries are eternally fascinating (to me !) . One thing I suspected, and AI2 sort of proved, is that Tesla's ability to model the real world is unrivalled ( nobody really mentions this rather important moat). That is a mix of software, mathematics, physics and engineering expertise, to model whatever they need to optimise - rather than rely on "expert opinion". What's interesting here is the "bias towards the proven result" - once you realise the reasons why side cooling is better - "Well it's obvious! "...(except it wasn't at the time...)
Amen!
And they said "it's not that big a deal!"
What do I feel like every other manufacturer is basically going to be like "make the cell bigger" and then panic when their fleet is on fire because the customer floored it or put it under real world load.
Ladies and gentlemen DUMP your money into Tesla. At least as long as Elon stays.
Small correction LEAF batteries aren't actively cooled, just passive heat transfer, 1rst gen Ioniq is example of active air cooled battery
I don't think I mentioned active or passively cooled. insideevs.com/news/482245/nissan-leaf-repair-liquid-cooling-benefits/
@@thelimitingfactor "The first generation Nissan Leaf was air cooled"...
All generations were convection cooled. Only the Nissan e-NV100 was fan air cooled.
With base plate cooling, the cans are attached to the cooling plate rather than a structural plate. I imagine the latter yields better structural strength/rigidity.
Great video. Shows the power of computer simulation ie gut feel is not always right.
Well done. Thanks Gordon.
Regarding Thermal function of the top/bottom Plates. My take is because your Tested 4680 Cell was not used under pressure, in a car, prevents us from seeing the Actual affect of heating cycles. Hope that @Munrolive will dissect one.
Imo, this all ties in with the "odd" charging curve displayed by the "Korean cousins".
They seem to have a "front half,/ rear half" temperature differential which they can't control.
Permission to use "smug face".
I'd reached all these conclusions (without calculation) 😁
.
Essentially "the best cooling is no heat production".
😁 Permission granted!
@@thelimitingfactor
😁😏
Here’s where Tesla’s “Fail Fast” philosophy makes a difference. Tesla’s engineers very likely just came up with several different ways to cool the cells, including ribbon cooling, and just measured the temp difference under load. They didn’t write a paper, they just found out what worked best via direct experimentation.
It is often the case that you can make something and tear it apart and remake a new design of it five times in the amount of time you could verify something theoretically.
In other words, building something and live testing it in the real world is often faster than theoretically modeling something before building it. Also, building things live gives opportunity to stumble upon new findings and knowledge that might well have been missed by modeling something to get a result before building it.
I worry a lot about todays engineers getting too far away from a hands on understanding of materials. It is better to have an idea and build it and test in the real world and refine and rebuild than it is to computer model something to death before ever building it. Sure there are exceptions such as when you need to know if something is possible at all, but hands on is the quickest and best teacher.
Surely this heat management advantage is going to play into improved charging characteristics. Either for one of charging or constant cycling such as the Tesla-Semi is likely to experience.
I definitely think that the thicker casing of the structural 4680 cells will help spread the cooling more evenly around the entire circumference of each cell, helping to negate the increased ratio of diameter to length found in the 4680 (as well as the structural benefits).
Thank you for all the work you do for the Tesla community 👍
This is amazing analysis. Rivian and Lucid used plate cooling on their EVs. I think that they will have performance issues when performing back to back drag races. Tesla Plaid does not have this issue since they fixed the serpentine ribbon cooling to a parallel ribbon cooling like the Model 3. The optimal design for heat dissipation then would be the tabbed 18650 cell with parallel ribbon cooling based on your analysis. Great job Jordan.
It is interesting that Lucid seems to not have overheating issues despite their bottom cooling design. They started out designing battery packs for EV racing so they must have some other innovation that results in better cooling.
Interesting.
Note that Peter Rawlinson threw side cooling under the bus in favour of Lucid's side cooling. His argument was that it is difficult to get reliable contact with the curved surface area of the battery. I was not convinced.
@@bru512 Peter is right, it is hard to get reliable contact. Tesla bought the company that was making the cooling tubes for them because quality control was not up to the required level. Basically there is a coating on the cooling tube that is cooked on, and it has to be cooked just right. Too "raw" and the surface is slippery and the glue that bonds the cell to the tube slips off. Too "well done" and the coating cracks and peels off the tube. Then the glue used has to be the right mix of the two ingredients, the applicator tube has to be properly maintained and changed often as it will clog over time, etc. One guy who was what they call a CTA (cell tube attach) worker blew off changing the application tips his whole shift, they had to bring finished battery packs back from Fremont and he got fired.
Yeah, just like the "experts" said, the Model 3 battery packs being made on an automated line was impossible, and when they say that just add "for anyone except Tesla" and you may have a true statement.
@@tribalypredisposed How do we know they don't have heating issues?
EV racing only requires packs to last a few cycles.
@@bru512 well, the cars are in the wild now and I don't think there have been reports of acceleration being limited due to heating. I may have missed them, of course.
One thing that may be helping Lucid avoid heating issues is that the pack is enormous at 118 kWh, so the acceleration heats each battery cell less because there are more of them.
Tesla's battery cells don't really need maximum thermal performance typically. Cells don't get that hot normally unless in extreme conditions like at the track perhaps. Also, uniformity is good when cooling or heating, you don't want the bottom of the cell to be different than the top half, so ribbon cooling cools the entire jacket which then cools the entire cell, making it uniform temperature. Good thermal management is needed for long term performance of the cells, but having maximum thermal conductivity is not even really necessary.
You get passive baseplate cooling as this up against the pack exterior. The only way to side cool is ribbon.
Nope. Base of pack for venting of liquid in event of thermal overload. Top of pack is via pink goo of death.
I am probably a bit late to mention that if doing base cooling, and there is a thermal runaway event, the expansion has to go down through the base, destroying the cooling at a time when it is needed most
Great overview! I hope someone can further clarify the "peak voltage power delivery" aspect of the 4680 vs 2170. Currently a Model 3 loses massive amounts of horsepower as battery SOC gets lower. I assume it's due to rapid voltage sag. Range may be unaffected but a Performance Model 3 with 30% SOC drives nothing like one with 85%. Will the 4680, by it's design, be more able to perform similar to the Tesla Plaid, where the car performs steadily until a much lower SOC is reached?
Lower resistance in the 4680 may help!
Power throttling is a factor of inverter design. Power = Voltage × Current this means you can maintain power even with dropping voltage by increasing current draw and the limiting factor for current draw is the inverter. The tabless design will increase max current with in the cell but I don't think Tesla is drawing max current from the cell.
You cannot see which exact cell is defective from a BMS perspective, you can only see that for each voltage set of cells. A faulty cell will drag down voltage of the whole set as they are wired in parallel. It will also limit how much that set can charge. To determine which specific cell in a set is defective, you have to disconnect the cells and test them individually.
Say your battery has 4,416 total cells arranged as 96s46p. If any one of those 96 serial sets of 46 cells in parallel is significantly different than the rest of the battery pack, the entire serial string capacity suffers. This is a disadvantage of small cells (e.g. 18650), as there are so many more of them increasing the likelihood that one will fail compared to 4680s. If an individual cell completely fails with a short - that is actually a better case. The fuse link will blow and you are just out 1/46 of the capacity of the battery as it is taken out of the entire circuit. Same thing if it fails with an open, although in that case no fuse will blow.
I'm trying to square that with the fact that in the examples I've seen, 1 cell has dramatically impacted pack level performance.
@@thelimitingfactor
AKA "The Porsche Problem"?
@@thelimitingfactor Yes, one cell can dramatically impact pack level performance. But you cannot electrically measure the difference between that cell and the rest of the cells in the same voltage set with a BMS. What you can see with the BMS voltage probes is a difference between that set voltage and all the others, as you showed in the video.
Great video as always, appreciate the humility at the beginning further adding to your credibility. Look forward to your next videos!
I wonder where the Munro and Associates would like to see the breakdown of your battery ideas
Great video man. I wonder if this will be the long term or shorter term cooling solution. I would imagine it would continue to evolve as new silicon doping and material changes with v2, v3, etc if they are indeed serious about million mile battery.
Also big congrats on closing in on 100k subs!!
Getting there slowly! lol
Things have kind of levelled off.
With higher amount of cycles for LFP + lower degradation due to better thermal management of the tables design these 4680 batteries will likely outlive other components of the car enabling reuse of battery packs in other cars and/or in future storage packs.
While simultaneously giving more power as the shorter travel path of electrons enables higher current flow increasing the acceleration possibilities and also the charging speed possibilities.
Looking forward to cars being charged at vastly shorter times without degrading them.
Will likely be even more important for Semi & cybertrucks than current passenger models.
Thank you for your very detailed in depth videos! This is braingasm for a chemical engineer 🤩🤓
You'd think! But as Drew Baglino has said, the batteries have already had issues with the cells outliving the pack. Battery packs are a difficult challenge from a durability point of view as well.
@@thelimitingfactor interesting! what part(s) of the packs are failing before the cells? Maybe you have already mentioned this in another video?
So now Im thinking that the extreme thickness of the material for the 4680 cans is not just for structural rigidity, but also to increase the thermal conductivity of the cans themselves.
I wonder if there's a way to use the natural void in the center of the cell for liquid cooling? This would allow the cells to be closer together creating a higher with volumetric energy density pack. Maybe you could even have a passive mechanical thermostat in the center of each cell, allowing each cell's temperature to be controlled as an individual.
You could, but it would create many other issues with manufacturing, cost, and weight.
I thought I heard ( but I am not sure from where) that with generation 2 batteries and the new 3 stages combined rolling equipment they were going to omit the copper current collector and laser weld the copper foil turnovers directly to the can base.
Next video
I would also have thought Tesla would switch to "bottom/top cooling" since it's far less complex to assemble. The only thing I had my doubts about was the height, because the cells are now even longer, which would make the floor structure difficult for a sedan.
Nice work!
A battery is not 95% efficient, the graph you are showing to this is the on board charger efficiency
A battery 21700 has 15mR milliOhm of resistance, with a current draw of 0.2-C in high-way driving conditions, this is 4500mAh*0.2 = 0.9 A per cell = 12mW loss for 5h = 60mWh lost, on 16.2 Wh total cell (3.6V, 4.5Ah) = 0.37% lost = 99.63% efficiency of battery cell only!
Battery is roughly 99.1-99.5% efficient if you include bus bar and cabling losses in battery pack assembly
When I spoke to battery scientists about it, I usually was advised of an efficiency at the battery level of 90-95%. The charging equipment was also showing in the image (also 95%).
@@thelimitingfactor that is true
newer battery chemistries since 2018 model 3, now have power NCA and NMC cell chemistries where the battery is designed more to efficiency, this leads to >97-98% battery efficiency at least in regular power levels. during supercharing this drops way in of course, due to the large currents and thus resistances that add up.
OBC charger is sometimes 93% some products 96%
while the SiC inverter of model 3 achieves ~98% efficiency, and newer versions even 99%
A very illuminating explanation... confirming also that Tesla's engineers considered all possible functions (thermal, electric, structural, ...) of every part in order to improove vehicle's efficiency...👍🤙
The way that the current collector jumps from the outer edge to the core, skipping the middle layers, is exactly what you would want to reduce the temp increase in the center. If you contacted all the layers then the outer layers would "use up" the cooling potential, make it less cool where it hits the center hotspot. There may be an analogous electrical advantage if the electrical resistance to the center is also higher, though not obvious why that would be so in the tabless design.
Sometimes the initial choice for a particular design maybe the best over time, even for a company which keeps refining its design. The side cooling ribbon design is still bringing the best result for Tesla. However Lucid is doing its cooling through the bottom end of it cells which does not make too much sense in terms of surface contact area!
Maybe this is a dumb question, but why packing thausen of cylindrical batteries vertically when you can stretch the battery to 2 meters long pipe like design, and packaging in a horizontal styles, like a floor made of pipes?
That's basically what byd did with the blade, except a ladder frame. Horses for courses.
Also, cooling between cells maximizes cooling while minimizing cooling lanes required. having coolant flow on the bottom or top of the pack would only cool the cells on one side, and need to be insulated on the other side. (you don`t wanna cool/heat the structure around the cells)
if the cooling channels are not the full height of the pack, they probably won`t hurt rigidity, because the lever is the highest on the top and bottom of the cells anyway. think of hollow structures or half-timbered structures.
Great point!
in summary, Peter Rollinson of Lucid was super wrong to have said cooling from the top was the best. I guess he needs an internship at tesla.
😁 I just watched his video again. He's mainly saying they chose baseplate cooling because they found ribbon cooling tricky.
@@thelimitingfactor Correct, that was my interpretation. Need to wait 10 years to see how long the lucid batteries last.
Hmm, could they ever do center cooling sense the middle of the can is hollow anyway? Im having trouble putting the idea together in my head but im sure someone has thought of it.
A few people have mentioned this. It would add too much complexity and weight for little benefit.
Thanks Jordan! Vinfast of Vietnam is starting to use an innovative battery pack per Sandy Munro.
🙌🙌🙌😀
Greta video but I would like to point out that plate cooling also utilizes the Can of the cell to cool the sides of the cell as well, Plate cooling allows the copper and aluminum conductors a less Thermally restricted path for heat transfer whereas in just side cooling 20 percent area the heat has to travel through multi-layers within the roll of the cell to cool. This is still effective but the difference is seen when the cells are stressed thermally in high-current modes of operation. This is why Ribbon cooling is not as effective in race conditions or long-duration high current draw situations. Of course variables such as coolant temp can factor into this but as a researcher who has experimented directly with cylindrical cells , Plate cooling a cell that is designed for it is more effective at heat transfer. I am curious as to why you did not use the Lucid module design ( a plate-cooled cylindrical cell ) in any of your comparisons as that would have been a more comparable difference between the 2 types of cooling methods.
What a surprise ,I was sure it would be done at the copper anode end of the sell, it goes to show how clever Tesla truly is, Thanks for your update .
Thanks a lot for the video and for sharing the reference. You can still use the explanation showed here in order to demonstrate that the prismatic can be as good as a cylindrical 🙂. The conductivity alone is pointless, and using the thermal resistance is more realistic for choosing the best cooling direction ( Thickness/(Conductivity*Cooling_surface). In the case of the cylindrical cell, the radical conductivity is quite poor compared to the height direction. But the radial direction has higher surface. There is a possibility to make the top/bottom cooling better but due to cell integration and important contact resistant in the height (gasket, gaps, electrical insulator) , the height is less interesting. Finally in the case of the LFP from CATL , you can apply the same reasonning and due to the use of the LPF you can have a higher Tmax overall and you do not need higher cooling surface. The sum of improvement of TESLA is driven by cost per performance and tooling is one of them. My guess is that there is less ivestment by staying in cylindrical instead of switching to another form factor. The increase of cell size has been done by CATL and BYD already in 2018 and they have also updated tabs and the number of JR inside the cell in order to cope with the thermal drawbacks
Sure thing! Depends on what your goal is. A full form factor comparison can't be done the comments here. I already point out an upcoming video that Qilin has better thermals than 4680 structural.
@@thelimitingfactor looking forward to it
I LOOOOOOVE that kind of content.
Pretty much nailed it. One of the reasons I have a hard time recommending cold plate cooling.
I believe the Tesla battery pack design consist of a group of cells connected in parallel and then these groups are connected in series. Within a paralleled group of cells the BMS doesn’t know anything about the individual cell, just about the groups. I’m not certain if this affects any of your conclusions in the video, though. 3:30
🤗THANKS JORDAN AND YOUR PATRONS 👍
GREAT EXPLANATIONS ,EASILY UNDERSTOOD BY LAYPEOPLE…🤔
It would be great to buy you lunch someday,since I am in N.E. OHIO…IF YOU COULD SPARE THE TIME 🤗💚💚💚
n March 2021, Optodot was issued U.S. Patent No. 10,950,837, titled “Methods of Producing Batteries Using Anode Metal Depositions Directly on Nanoporous Separators.” Third Generation NPORE® ECS Electrode Coated Separator technology, developed with funding by the DOE, aims to reduce the cost of manufacturing lithium-ion batteries and the inactive components cost by 20-40%, while improving battery safety, lifetime, and energy and power density. NPORE® ECS incorporates new inactive components of separator, current collectors, and termination materials, and utilizes a simpler and faster battery assembly process. With ECS, electrodes-for example, lithium metal anodes-are directly deposited onto the separator to form a separator/electrode stack.
I wonder if the cooling becomes less effective towards the rear of the Pack as the heat is absorbed from each cell?
Does this mean the cells towards the rear of the Pack will degrade faster overtime
Or will Tesla reverse the flow of coolant to prevent this from happening
Does this mean the cells in the centre of the Pack will degrade quicker than the ones on the outside of the pack? or are the cooling channels routed differently within the ribbon to compensate for this
No, this is why I said in the video that the gradient isn't as large as you'd expect.
US Patent No. 11,387,521, issued July 12, 2022, is directed to an advanced nanoporous separator for lithium-ion batteries which has far superior safety properties such as better heat conductivity and non-shrinkage compared to conventional separators currently used in lithium-ion batteries. US Patent No. 11,387,523, issued July 12, 2022, is directed to a new generation lithium-ion battery in which the cathode is directly coated onto the separator. This technology reduces the complexity and expense of the equipment used to fabricate the battery, for flat and prismatic batteries that can be used, for example, in electric vehicles. This US patent also enables alternative manufacturing processes, such as where a non-dried, wet cathode is applied to the separator.
I would have also gone for a top/bottom cooling. In a uni e-motorcycle competition i designed the battery pack cooling i such a way, not appreciating the full jellyroll + can approach since the infor the papers showed the later years was one approaching the jellyroll alone, with very poor radial conductivity.
Also, we weren't focused on durability buy max T in 5-10 battery cycles.
For the new prismatic cells bottom heat sink utilization was a no brainer
Cylindrical 2170 cells have 10x better cooling axially compared to radially. Tabless improves that significantly. Plate cooling has a similar thermal performance as ribbon cooling up to the can but ribbon cooling performs better at spreading that heat around both bottom and top of the cell better.(area is less for plate cooling but epoxy gap is smaller and more consistent so overall thermal interface is about the same. Tesla isn't able to get the ribbons to form exactly to the shape of the 21700 cell. 4680 is a huge improvement on many fronts. :)
Incredibly informative and easy to understand. I'm not a chemistry geek, so there are times when the explanation flew over my head. But overall, it confirms my faith in the 4680 battery, especially if the Tesla owner plans on keeping the car for a long time
In summary, "Ribbon cooling provides greater surface area, which results in higher efficiency of cooling". However, increasing cell size 5x also increases thermal mass. Meaning the center of the cell is cooled exponentially less than the outer layers of the cell. In theory, tabless technology coupled with ribbon cooling applied to smaller cells like 2170 & 18650 would be the most efficient way to maintain thermal management of the cell. What this means to me is that thermal cooling for the cell/pack is not Tesla's priority. The more important variables to consider, which Elon emphasizes consistently, is cost and scalability. I think 4680 would not have been possible without the tabless innovation and its impact on thermal management. Tabless cells will prove to be the single greatest innovation in batteries due to its low resistivity = less heat generated and the increased rate of cell fabrication. The next innovation will be the solution to high energy dense cells using silicon. Exciting times we live in!
Awesome breakdown, thank you. Here in southern Appalachia we heat exchange with our stills the same way, copper tubing with max thermal exchange through the walls of the tube, not the ends. ... . .
None of the Nissan Leaf's, including the second generation, have active thermal management. This is their Achilles heal, as it causes their packs to last significantly fewer miles than other EV's, and Nissan is obtuse regarding this subject.
Second time watching through. Takes awhile to get all of it and I'm sure I'm not there yet. BTW, could we please have some color choices for the hats. Please.
Not sure why they stopped at 46. They could have gone for 8080 with a hollow core instead. This would give them cooling lines through the whole core and 80% of the circumference. It would also make it trivial to connect those in series, as they line up that way automatically when threaded onto the central cooling pipe.
Can you do an analysis on StoreDot and its battery chemistry? Can you confirm(debunk) the validity & effectiveness of Silicon based anode NMC cathode? Is it a sham or is solid state truly be the future of battery innovation?
If it can conduct electricity, it can conduct heat. The current collectors at the top are wicking away heat then conducting over to side can. Dual purpose for sure. The tabless design is also drawing heat out from the center for same reason ohmic resistance is lowered. Did not surprise me that side cooling works.
If I was rich, I'd donate ALOT lol I want you to keep making these for a LOOOOONG time 😈😇
Do you know of any studies showing which type of batteries are lighting up and what the conditions were? We've seen plenty of fires in line but nothing about which ones, if it was design problems or external. Thanks for your work.
No! I'm curious about that as well though.
Yeah, especially Porsche - specialised cells. Not known to focus on safety! Shouldn’t be when vehicles are idle, parked, on ocean.
Learning is so great, thanks for bringing us all along on your journey!
Question: the cells are encased in some tough polymer as the Munro tear down showed. I get that the polymer has limited cooling capacity, unless Tesla designed that material to absorb some heat. Once warmed maybe any potential cooling bennies are negligible?
Nice job! I think you may have overestimated the thermal conductivity along a thin sheet of electrode and underestimated the thermal conductivity through a thin plastic sheet. It would be interesting to see if the people doin the modelling published numbers for that.
Thermal management is all for fast charging.
I don’t plan to fast charge my Tesla almost at all. I could wait a longer time I fast charge or could fast charge just the minimal amount.
I built a LFP battery for a RV that gets used daily - 6yrs old now - new cells were Bottom balanced @2.75 volt -
cell balancer units were not reliable at the time so did not install one - battery discharge to 40% daily - solar charge -
I check battery cell balance every few weeks - cells hover around 4mv difference - Amazingly Good
Ribbon cooling is a little bit cheaper to implement in manufacturing. I'm pretty sure that's the only reason because the electrode itself is actually a long flat ribbon that's been coiled up which means that cooling the side is not actually cooling the side, you're cooling the ends essentially. The heat actually can travel faster along the width of the ribbon up to the top of the cylinder and then out to the sides because of the larger thermal mass then it does around the coil a bunch of times to the outside. Cooling the outside of the cylinder is easier than cooling the top and the bottom together and cooling the side cools it roughly equally. It's pretty good, it's pretty cheap, it's pretty easy to do, pretty reliable, and overall just a good balance of lots of factors. It's not really the best in any regard but without spending ridiculous money this is kind of the best compromise for the money
Given that single cell failures have a disproportionately large whole of vehicle cost, why is per-cell monitoring and isolation not built in? I can imagine that a cell absence would affect battery bank voltage and current characteristics, but are there not simple electrical solutions to those?
Excellent analysis. What you did not say is that one end of the battery is great for cooling, but at the other end there is the electrical connections for both positive and negative. Perhaps also some thin wires which monitor the performance of each cell. So, it will both add complexity and be of less value thermally, perhaps not worth the effort, to extract heat from both ends of the battery in the end cooling scenario.
Eh, I'm surprised why engineers haven't just enclosed top and bottom of cells to make it water proof, pack batteries closer so body of the cell creates channels, directly cooling cells and in case of thermal runaway of single cell you have A LOT water right next to cell to boil off before endangering nearby cells
also, the bottom of the cell has a noticeable airgap between the active materials and the casing, thus really not effective cooling area Vs side. Cooling from the top gets challenging as that's where the electrical connections are made for positive and negative welds.
Bigger batteries with better cooling mean that you can potentially charge every cell on its own, and potentially less cells can be bunched into a group/pack, the ideal being each cell a pack.
I remember when Rivian engineers were making funny remarks about Tesla's ribbon cooling.
There often discrepancies between theory and practice. Only by testing can one know for sure and it would appear that Tesla has left nothing to chance and chosen the optimal cooling strategy.
It would seem that the pack encapsulation by the pink material would contribute to heat removal if it is thermally loaded.
The model 3 LFP pack is already using large format prismatic cells, Munro showed one in one of it videos of him visiting a vendor. I couldn't find any other sources to confirm it, nor did they go into any significant detail like how it was cooled iirc.
Let me add to the rationale in this video.
(1) The anode foil wick heat more effectively towards the base plate only to be impede by the electric conducting wafer (a heat insulator) leaving a high thermal gradient between anode foil and base plate.
Remedy? Solder all anode flaps at the base plate which will not set back electric current.
(2) given no soldering option, heat transfer must use radial path towards cylinder wall. Heat transfer radially through chemical and insulation is undesirable but it benefits from a higher wall area as opposed to base plate with a wafer heat insulation. With better coolant circulation we can pull it off, heat in the core.
(3) per base plate cooling architecture - only a fraction number of cells (cylinder wall and plate) can be directly connected to the chassis to benefit a direct heat path to a metal structural plane with coolant channel under breath. The rest are (electrically insulated) that impedes heat flow out of the base plate. Heat at the base plate must migrate around to cylinder wall to be wicked off.
Hence the heat sink via cylinder wall become a prevalent architecture.
I note there's construction at Giga Nevada! (Interesting! 😉)
Class is in session! Let's go Jordan!
I had to keep pausing and rewinding ! There's a ton of info in this video ..
Yes clearly the correct approach in this case though it's not perhaps the case for other form factors.
Can you do an analysis on StoreDot and its battery chemistry?
Lucid battery pack video stated that it is difficult to make the ribbon channel to be in full contact with the cylindrical cells in mass production. Instead of 20% of the area as you stated, you may get 1/4 of 1/3 of it if the ribbon is misaligned with the cell cylinder. How did Tesla solve this issue? Maybe worth to do a video on that.
Conductive adhesive and precision engineeering
Thanks again for the great content :) Love your deep dives :)
Why arent cells built and cooled internally. Imagine a copper or alumium pipe of diameter 6mm with 1mm walls (or whatever size is appropriate) and the active battery materal is wound around this pipe. You could build extrmely long bayteries. Rather than 46-80 maybe even 50-800 or even bigger
And then you just run your cooling internally
Because it's all about surface area. Cooling the core without cooling the outside would cause more problems. Plus, it would be more expensive.
I believe your original assumption concerning end verses side cooling to be correct when it comes to end/tab cooling, though side cooling could potentially be superior if implemented correctly. The problem as I see it is that Tesla’s ribbon cooling solution cannot achieve equal cooling to all the cells, meaning the cells at the intake side will experience lower coolant temperatures compared to those cells at the outtake side of the pack. The other problem is that the ribbon as you have explained is only contacting at best 20% of the cells side area, a further problem is how to guarantee adequate thermal contact between the cooling ribbon and the cell side cylinder wall. No matter what, you have given me something to think further about.
I've looked into that, very little temperature variance between the intake/outtake is a small fraction of a degree. Also, that was perfected even more with the intra-ribbon loops.
If a pack has faulty cells, can Tesla disable and isolate part of the pack? Could this explain the weight of the Austin Model Y 4680 structural battery pack as per Cleanerwatt's analysis?
No, the Tesla pack would be limited by limiting the voltage range (if it is)
Great cooling video: Question: Why have they not run the cooling in a more serpentine fashion to create a greater surface area for cooling?? ~~~~~~~~.. One cooling arm could cool two row of batteries with more than 50% of the can surface area in contact. I realize that it would add a small amount of weight, but this seems negligible compared to the battery life savings with improved thermal management.😀 Thanks
That's a million dollar question! We don't know their current baselines. I'm assuming they made the ribbon decision design on balance and it could be an area of future improvement.
will the next model of tesla include a toaster in the console?
Ty for taking the time to so clearly and professionally lay out this presentation. The graphics and production quality hit way above what you'd expect given your subscription count. I know this took a long time e to put together. I learned a lot. I hope your channel grows. Subbed.
Thanks David! Yeah, they're a lot of work, lol.