Tesla’s Battery Day presentation outlined big improvements in cost, energy, charging time, and pack size. We will show you our estimate of the new charging profile, an overlay of 2170 vs 4680 pack size, and an estimated cross-section of the new 4680 pack. We’ve also estimated the series-parallel cell configuration and a potential way Tesla will electrically connect the cells.
In a previous article (ref), we outlined the results of a preliminary thermal analysis that suggested that Tesla’s new 4680 pack would use a much simpler and easier to assemble flat plate cooling scheme and ditch the cooling snake in between the cells that they have been using since day one.
While Tesla didn’t specifically say it will go with flat plate cooling, it is strongly suggested for a number of reasons. The primary one being that you can’t effectively cool these larger diameter cells through the sides. The most effective way to cool the new tabless design is through the ends of the cells since the new tabless design provides a great thermal path between the cell cap and the inside of the cell.
The copper anode electrode plates themselves act as great cooling fins and provide nice uniform temperature distribution within the cell. The other reason we think flat plate cooling is likely is that Elon Musk said that the cells are glued to the bottom and top sheets of the pack. While his quote is in reference to added shear strength in the pack, we think the top and bottom sheets will also double as cooling plates. In the Model 3 pack, the cells are glued to the cooling snake.
Our analysis makes a direct comparison between the Model 3/Y 2170 pack and a theoretical 4680 pack. You may ask why we chose the Model 3 pack to make the comparison when Tesla has stated that the 4680 pack is primarily aimed at the Semi and Cybertruck and Model S Plaid. The reason we chose to make the comparison with the Model 3 pack is that we know a lot about the Model 3 pack, so it’s much easier to quantify the improvements against a known baseline than an unknown baseline in the case of the Semi and Cybertruck.
At 85 deg F ambient, calculated charging time from 10% to 80% with the 4680 pack was reduced from 25 minutes to 15 minutes. If you only need a 50% charge, you can be done in 7 minutes. That’s almost as fast as gas.
The estimated maximum charging rate increased from 250 kW with 2170 pack to 275 kW with the 4680 pack. The 275 kW charging rate holds constant from 10% to 50% state of charge where we begin to taper the charging rate. We begin to taper when the cells reach their temperature limit assumed to be 45 degrees C (113 degrees F) based on Model X data. The taper point is a function of ambient temperature since the A/C system’s performance decreases at higher ambient temperatures. The charging profile comparison is shown in figure 1 vs time and vs state of charge (SOC) in figure 2.
Above: A comparison between the 2170 pack charging profile and the estimated 4680 pack (time-based), credit Keith Ritter PE Engineered Compliance.
Above: A comparison between the 2170 pack charging profile and the estimated 4680 pack (SOC based), credit Keith Ritter PE Engineered Compliance.
We see a significant reduction in the size of the pack
Eliminating the cooling snake allows Tesla to pack the cells closer together. This is a plus because it reduces the polar moment of inertia and improves vehicle handling since the mass is concentrated more towards the center of the vehicle.
Above: 96S9P 76 kWh 4680 pack Credit Keith Ritter PE Engineered Compliance
Above: Overlay of the estimated 76 kWh 4680 pack on top of the 2170 Model 3 pack, Credit Keith Ritter PE, Engineered Compliance
Cross section of our estimated 4680 pack shows top and bottom flat plate cooling and cell electrical connections
We see a much simpler way of making the electrical connections in our estimated 4680 pack. We think Tesla will eliminate the "finger collector" concept with wire-welds etc. and go to a simple plate-based collector that can be directly welded to either the "tabless" anode or the cathode-connected can. To connect the + of one group to the - of the other, they go back to their Model S-based "Inverting" technique where they flip the cells of every other parallel cell group.
Above: Cross-section of estimated 4680 pack credit Keith Ritter PE Engineered Compliance
As previously stated, we estimate both a top cooling plate and a bottom cooling plate with the cell glued to the plate. The cooling plates would also provide shear strength to the pack. We estimate a 30%/70% split in heat transfer between the cathode (aluminum) end of the cell and the anode (copper) end of the cell.
Could Tesla get by with just a bottom plate? We think not. This issue is discussed in further depth in the detailed discussion below.
Our thermal electrical model includes all the specifics of the pack. Dimensions, cell dimension, amp-hour ratings, thermal conductivities, etc. One key ingredient is the resistance of the cell since this determines the cell heat generation (I squared R losses). Sean Mitchell did an excellent interview of Ravi Kempaiah wherein Ravi explained the resistance reduction for the new 4680 tabless cells. The presentation is linked above and the resistance discussion is at 8:11 in the video. Based on his presentation we used the 4680 initial internal resistance = 3 mOhms @ 10% SOC and tapering to 2 mOhms at 80% SOC. The 2170 cell was 23/20/20 mOhms. So we reduced the cell resistance by a factor of 10. This compares to a 5-20 times reduction that Tesla quoted in its patent app.
In addition, the refrigeration capacity is an input to the model. We chose to stick with the existing Model 3 refrigeration capacity of 2-3 tons (depending on ambient temperature). However, it is possible Tesla might increase its stack chiller and AC compressor sizes to provide better pack cooling.
The thermal mass of the pack also comes into play. During the first part of the charge at the maximum charge rate, the cell generates more heat than the refrigeration system can keep up with, so the pack begins to heat up. The thermal mass delays the onset of cell overtemperature as it stores the heat. When the cells reach their temperature limit (45C=113 F), we begin to taper the charge rate.
Figure 6 gives you an idea of how the heat rejection of the pack and the heat removal capability of the A/C system compare. Note the decreased A/C performance at higher ambient temperatures. This is why the initial taper point off of max charging power occurs at a lower SOC at higher ambient temperatures.
Above: A/C performance vs ambient temperature, credit Keith Ritter PE
Can Tesla get by with only a bottom cooling plate?
It would be simpler and lower cost to have only a bottom plate. Most (70%) of the heat goes out the copper (anode) ends of the cell. Why not orient all the parallel cell groups anode end down and just use a bottom plate?
The patent is pretty clear that only the copper current collector will get the "tabless" end treatment. The aluminum collector still only has a single tab at the middle-end of the jelly roll that connects to the can. It likely just gets too complicated to do crazy interleafing of segmented tabs at both ends at the same time.
The heat transfer calculations say the bottom-cooling-to-copper-only configuration could only do 70% of what top/bottom plates can do and would create a larger axial heat gradient. The negative tab terminal end will be the hot spot. If we put a second plate on, that provides cooling to the - tab can end, we shift the hot spot about 1/3 the way down axially and can extract 30% more heat at the same "hot spot" cell peak temperature.
Two plates mean less cell thermal stress/deterioration and better cell life. Tesla will be gluing in these cells for life: The million-mile-battery. If it will also be ramping up charging kW some more to stay "leading edge," it HAS to do everything to ensure minimal thermal stress/degradation on these cells. A second cooling plate is cheap insurance. We think flipping the cells with every cell group also helps cell-to-cell conduction to spread-out the heat in the cells' axial and reduce thermal gradients between "top" and "bottom."
Sandy Munro, in one of his interview videos, also predicts there will be both top and bottom plates. He's been spot-on about his forecasts for Tesla's battery and structural changes so far.
*Thermal analysis credit:
Keith Ritter PE