Battery Cell Cooling Explained — Distributed Conduction & Anisotropic Cells

A cell doesn't heat up from a single point

The intuitive way to think about a cell's internal thermal resistance is "heat travels from the middle to the surface" — a single point source at the core, conducting outward. That picture is wrong in a way that matters: a cell's I²R heating happens throughout its volume, not at one spot. Every layer of the jellyroll or electrode stack generates its own share of the heat, and the layers near the surface have a much shorter trip out than the ones at the centre. Solve the conduction equation properly for that distributed picture and the resistance from the hottest interior point to the cooled surface comes out lower than the naive "point source" formula would suggest — by a factor of 2 if you're cooling from one face, and by a factor of 8 if you're cooling from both. Get this wrong and you'll over-predict the cell's temperature rise by as much as 8×, which is exactly the kind of error worth building a calculator to avoid.

Every cell format has a grain — and a natural cooling face

Both cylindrical jellyrolls and stacked prismatic/pouch cells are built from many thin layers, and that layering makes them anisotropic: they conduct heat much better in one direction than another. A cylindrical cell's metal current-collector foils run the full length of the winding axis, so heat moves easily along that axis but has to fight through dozens of low-conductivity electrode/separator interfaces to move across the winding (radially). A prismatic or pouch cell is the same idea in a different shape: heat moves easily along the flat electrode sheets (in-plane) but poorly through the stack (through-thickness). That's why real packs cool cylindrical cells from the can wall (the short, radial path) and prismatic/pouch cells from their large flat faces (the short, through-thickness path) — not end-to-end through the cell's long axis. This calculator lets you pick any cooling face, but it'll warn you when you've picked the slow one.

Why the internal cell resistance is often the bottleneck, and what this model doesn't capture

The genuinely uncertain part of this physics, why conduction — not convection — is usually what limits a battery pack's cooling, and the honest limits of a first-pass model.

The part of this physics that's still genuinely unsettled

A cylindrical cell's radial thermal conductivity — the one that matters for the recommended side-cooling path — is reasonably well agreed upon in the published literature, in the range of roughly 0.2 to 0.5 W/m·K. Its axial conductivity is not: different measurement methods in the literature report anywhere from about 2 to 30 W/m·K for what should be the same physical property, likely because the axial direction is hard to isolate experimentally from the rest of the cell's structure. This calculator defaults toward the higher, more commonly cited value, but discloses the uncertainty rather than hiding it — and conveniently, it barely matters for the recommended (side) cooling path anyway, since that path doesn't use the axial value at all.

Why conduction usually wins the argument over convection

It's tempting to assume that pumping more coolant or picking a fancier thermal interface material is the lever that matters most. Run the numbers on a typical cylindrical cell, though, and the cell's own internal conduction resistance is often the largest term in the stack — bigger than the TIM, bigger than the convection resistance to a well-designed liquid cooling loop. That's a direct consequence of how low radial jellyroll conductivity is. This calculator surfaces which term dominates for your specific case, because it changes where engineering effort is best spent: if conduction dominates, a better coolant or thicker cold plate won't help much — the fix has to come from the cell's own construction or from a shorter conduction path (better cell orientation, more contact area, side- rather than base-cooling).

What this model doesn't capture

The default heat-generation term is Joule (I²R) heating only. Real cells also have a smaller reversible/entropic heat term — tied to how the cell's open-circuit voltage shifts with temperature — that's typically only 10-15% of the Joule term at the discharge rates a cooling system is usually sized against, but it can flip sign with state of charge and between charging and discharging. It's left out of the default result and offered as an optional add-on rather than silently folded in. The conduction model itself is also 1-D: it doesn't capture 2-D/3-D spreading effects, tab/terminal geometry, or how heat behaves in a full multi-cell module with neighbour-to-neighbour interactions. Treat this as a first-pass sizing tool — the right order of magnitude and the right qualitative story (which term dominates, which face to cool from) — and verify a final design against a cell datasheet, FEA, or measurement.

← All guides · Open the Battery Cell Cooling Calculator