Cooling the tab or the body?

Here i found a video showing the difference between 2 cooling. The first by the body and the other by the tabs.
What is the best way to cool a battery pack? To thermally control it?

My first thinkingg was by the body, more contact surface for temperature exchange,(like tesla) but after listenned this video that making sense to me.. i'm not sure anymore...

Did someone already run some similar test ?? What is your result??

Here the video:
https://www.youtube.com/watch?v=_jd8REVB-c8

I was heavily persuaded as to the tab-cooling method for pouch cells by the attached research paper claiming ~3x better cooling (and reduction in cell ageing) . It seems the video you posted is arguing much the same thing.

View attachment LionTabVSSurfaceCooling.pdf

It makes intuitive sense when you think about a cell as being ~100 layers of material, much of it plastic, which is a poor thermal insulator. The thermal conductivity through all these layers is a lot worse than along them, where you have conductors running up to the tabs.

I would love to hear the thoughts of someone concerned with pack cooling for OEM BEVs or grid-scale BESS on the subject.

Tab cooling us cheaper and simpler to produce. That is why you see it in the model 3.
But it is piss poor in dealing with heat spikes and higher heat loads like having acrive heating to 40+C and then hammering the battery. Then it needs to cool and the thermal mass is much slower then pumping water directly over the center of the cell compaired to top only heating and cooling.
Advantage is that you cool the contact point thst have higher resistance so you are activly cooling the hottest part of the cell by default but the mottom part is SOL.

I expect a modified version of this system in the roadster battery. I expect it to have the same setup but with the top battery inverted. It would be stupid to have 2 cooling loops. And as it has dual batteries it has zero issues delivering 3000+ amps, wich is mich more then it can actually use.