Lithium-Ion Battery Pack Cold Plate Conjugate Heat Transfer

An EV powertrain developer needed to validate the thermal management strategy for a lithium-ion battery pack cooled by three serpentine cold plates, under the severe transient loading of repeated high-power discharge cycles. The core engineering question was whether every cell in the pack could stay within a safe thermal margin during a high-current pulse, and whether the coolant-side cold plate design was doing its job without unnecessary conservatism.

Answering that meant coupling the fluid behaviour of the coolant to the solid thermal response of the cells and cold plate structure in a single conjugate heat transfer (CHT) simulation — a much heavier computational problem than a fluid-only or solid-only analysis, which shaped every decision in how the model was built and post-processed.

With series-connected cells across the cold plates, a full-fidelity CAD representation of every cell, tab, busbar, and coolant channel would have produced a mesh far too large to solve for a transient CHT case in a practical timeframe. The CAD model was therefore deliberately simplified: geometric features with negligible thermal influence were suppressed or merged, cell arrays were represented with consistent, repeatable geometry to keep the mesh structured and predictable, and the serpentine coolant channels were preserved at full geometric fidelity since flow path and channel cross-section directly govern the convective heat transfer coefficient driving the whole analysis. This selective simplification kept the model solvable without discarding the physics that actually mattered to the result.

The CHT setup itself coupled a k–ω SST turbulence model for the coolant flow through the serpentine channels with conduction through the solid cell and cold plate domains, allowing heat generated internally by each cell during the discharge pulse to propagate through the pack structure and into the coolant simultaneously, rather than approximating the two domains separately. Cell-level heat generation was applied as a transient internal source synchronised to the discharge profile, so the thermal response captured not just a steady-state condition but the transient temperature rise during and immediately after the high-power event.

Because the simulation produced a large transient, three-dimensional temperature field across all 120 cells, a dedicated Python post-processing pipeline was built to extract and organise the results in a form useful for engineering judgement rather than raw solver output. The pipeline pulled per-cell temperature histories directly from the solved field, computed each cell's margin against its thermal limit at every time step, and flagged the governing worst-case cell and moment automatically — turning what would otherwise be a manual, error-prone review of hundreds of thousands of data points into a repeatable, auditable margin analysis that could be re-run identically for design iterations.