Cooling Plate Design - Battery Design

A look at cooling plate design and some of the example designs, circuits and hopefully some posts looking at the CFD.

An encapsulated cooling fluid that is circulated to the battery where heat is transfered to and from the fluid. Heat is removed and added to this fluid away from the battery pack using a radiator and/or heat exchanger.

Probably the most common battery cooling system used in electrified vehicles as the system can use water-glycol as the cooling fluid.

Examples: Porsche Taycan

heat transfer coefficient is typically 1000 to 5000 W/m².K

Designing an effective liquid cooling plate requires a multidisciplinary approach that balances thermodynamics, fluid dynamics, material science, and manufacturability. The key aspects to consider include:

1. Thermal Load and System Requirements

Heat Flux and Total Thermal Design Power (TDP): You must define the maximum heat load in watts and the concentration of that heat (heat flux in W/cm²).
Temperature Constraints: Identify the maximum allowable temperature of the battery cells and components. Use this to determine thea the inlet temperature of the coolant supplied by the system versus flow rate and system operation as a whole.
- cooling fluid temperature increases with heat rejected from a cell, leading to higher delta temperatures on wide spans of the cooling plate
  - Cell to Cell Temperature Gradient – in a battery pack this will limit the performance and lifetime. However, there will always be a cell temperature gradient and a cell to cell temperature gradient. It is very difficult not to have if we are rejecting or absorbing heat.

Sizing the coolant flow rate

Once the heat load is known, the flow rate needed to carry it away follows directly from the coolant’s specific heat capacity. The heat a pack generates at a given charge or discharge rate can be estimated from its internal resistance and current – roughly, P_heat = I² × R_pack, where the current I comes from the pack’s capacity in Ah and the C-rate being applied, and R_pack is the pack’s DC internal resistance (see DCIR of a Cell). At higher C-rates this resistive loss dominates the heat load; at low C-rates entropic and side-reaction heat become relatively more significant, so the resistive estimate is a sizing floor rather than an exact figure.

The required coolant mass flow rate is then:

ṁ = P_heat / (c_p × ΔT_coolant)

where ṁ is the coolant mass flow rate (kg/s), c_p is the coolant’s specific heat capacity (roughly 3,300 J/kg·K for a typical 50/50 water-glycol mix, versus about 4,180 J/kg·K for pure water), and ΔT_coolant is the temperature rise you are prepared to allow between coolant inlet and outlet – tighter cell-to-cell temperature gradient targets push this number down and the required flow rate up.

Worked example: a 75 kWh pack (~208 Ah at 360 V nominal) fast-charging at 2C draws roughly 416 A. With a pack DC internal resistance of 60 mΩ, that gives P_heat ≈ 416² × 0.06 ≈ 10.4 kW. Allowing a 5 °C coolant temperature rise with a 50/50 water-glycol coolant (c_p ≈ 3,300 J/kg·K) gives a required mass flow rate of roughly 10,400 / (3,300 × 5) ≈ 0.63 kg/s, or about 36 litres/minute for a coolant density near 1,050 kg/m³.
This is a sizing estimate, not a substitute for CFD or bench validation: it ignores the coolant-side pressure drop through the channel geometry (see Internal Flow Channel Architecture below), pump curve matching, and the transient thermal mass of the plate and cells, all of which affect whether the calculated flow rate is actually achievable and sufficient in practice.

The Kia Niro/Hyundai Kona use cooling plates and a liquid coolant fluid. These plates cool the lower edges of the pouch cells that are arranged in 5 large modules and hence 5 cooling plates.

The two stacked modules at the rear of the pack appear to be fed from the two outer coolant plates in series.

The concern is that these two rear modules will see higher coolant temperatures than the 3 modules that run the length of the pack.

The two stacked rear modules are likely to age faster.

Thermal Conduction and Temperature Gradients – whatever the cell format and however we cool it there will be a temperature gradient. This is described by Fourier’s Law of Conduction. An empirical law stating that the rate of heat flow is directly proportional to the temperature gradient and the area of heat transfer, and inversely proportional to the thickness of the material.

Porsche Cayenne EV: Liquid cooling plates are arranged on the top and bottom of each module.

The cells appear to interface directly via Thermal Interface Material (TIM) with each plate (see module image below).

Cooling on both long edges of each cell will halve the top to bottom temperature delta in a cell.

Hydraulic Limits: Determine the available coolant flow rate (e.g., litres per minute) and establish a strict pressure drop budget. Exceeding the pressure drop limit will overwhelm system pumps and increase parasitic energy consumption.

2. Internal Flow Channel Architecture

Topology Selection: The internal geometry determines how effectively heat is extracted. Options range from parallel channels (low pressure drop but risk of uneven distribution), serpentine channels (good uniformity but higher pressure drop), to advanced micro-channels (essential for heat fluxes >50 W/cm² but highly sensitive to clogging and pressure drops).
Channel Cross-Section and Hydraulic Diameter: for a given topology, shrinking the channel’s cross-sectional area increases coolant velocity and the convective heat transfer coefficient, but also increases pressure drop roughly with the square of velocity – the two have to be traded off against the pump’s pressure budget. Non-circular channels (rectangular slots are typical in a machined or brazed plate) are usually sized using the hydraulic diameter D_h = 4A / P_wetted, where A is the cross-sectional area and P_wetted is the wetted perimeter, so that standard circular-pipe correlations for heat transfer and pressure drop can still be applied. Loop or “race-track” layouts – a single channel that winds out to the plate’s edge and back, as used on the Ford Mach-E plate pictured below – behave hydraulically like a long serpentine channel: one continuous flow path, so the cross-section has to be sized for the full plate-length pressure drop rather than the shorter run seen by any one branch of a parallel design.
Surface Enhancements: To maximise the convective heat transfer coefficient, designs often incorporate pin-fins, dimples or offset-strip fins (OSF). These structures deliberately induce turbulence and force the fluid’s thermal boundary layer to periodically break and restart, vastly improving heat absorption compared to continuous straight channels.

Ford Mach-E: The water enters at the centre on one side of the plate and then flows around the “race track” and back via each side to the central outlet. The dimples manage the structural offset of the two halves of the plates and break up the liquid flow.

The pressure drop offered by this type of a design is less compared to a standard S bend for the same channel height.
The density of these dimples are adjusted to spread the flow towards the edges of the plate from the centre and also to locally increase the fluid velocity by decreasing the sectional area.
The person in the video was talking about turbulent flow. When a flow exits into a large cross sectional area such as this, the flow is mainly laminar.
Not much information is available on the distance between the top plate and the bottom plate. This is an important factor that decides the pressure drop and the heat transfer coefficient. However, this design heavily relies on the large surface area for convective heat transfer rather having a high heat transfer coefficient.
The top plate is used to reduce the temperature difference between the inlet and outlet of the coolant due to its higher conductivity.

Flow Uniformity: The design must prevent “dead zones” where fluid stagnates, causing localized hot spots, and ensure balanced flow distribution across all channels.

A question the battery thermal team often face is: What is the Optimum Design of a Cooling Plate for Uniform Cooling? To achieve optimum uniform cooling, a cold plate must be designed to minimize temperature gradients across its surface and prevent localized hotspots.

3. Material Selection and Galvanic Corrosion

Thermal Conductivity vs. Weight:
- Copper is the gold standard for high-performance applications due to its superior thermal conductivity (~400 W/m·K), though it is heavy and expensive – hence unlikely to see it’s use in any battery pack other than very specialist applications.
- Aluminium is widely used for weight-sensitive and cost-constrained projects despite its lower conductivity (~205 W/m·K).
Material Compatibility: You must strictly avoid mixing dissimilar metals (like bare aluminum and copper) within the same cooling loop to prevent galvanic corrosion. If mixed metals are unavoidable, specialized corrosion inhibitors must be formulated into the coolant.
Electrical Isolation: typically made from thin aluminium for low weight and high thermal conductivity, means they need to be electrically isolated from the cells.

4. Coolant Chemistry

The choice of coolant—such as deionized water, water-glycol mixtures (ethylene or propylene glycol), or dielectric fluids—dictates the thermal capacity and viscosity of the system.
Additives and inhibitors are crucial for preventing scaling, maintaining pH, and protecting the metal surfaces from oxidation and microbial growth over the component’s lifespan.

5. Manufacturing and Structural Integrity

Manufacturing Method: The fabrication technique must align with the channel complexity and production volume. Options include CNC machining for prototypes, Friction Stir Welding (FSW) for highly reliable, leak-proof aluminum plates at volume, Vacuum Brazing for intricate aluminium and copper micro-channel assemblies, and 3D printing (Additive Manufacturing) for complex, bionic flow paths.
Leak Prevention: The cold plate must maintain hermetic sealing under operational pressures. This requires robust joint construction (like solid-state FSW or vacuum brazing) and reliable O-ring gland designs to handle thermal expansion and pressure spikes without extruding or leaking.
- Coolant liquid can leak from plates, hoses and most likely joints thus risking electrical isolation.

VW ID4: The cells are inside an aluminium module case, these are then in an aluminium enclosure and the cooling plate (Heat Sink) fits under the enclosure. This keeps all of the coolant fluids away from the inside of the enclosure. The coolant connections, these are connected to the outside pack frame and at the opposite end to the HV and control connections.

Image [Flying Tools] shows the cooling plate of an ID3/ID4 battery pack. This particular pack has had some damage to the underside and cooling channels.

If the pack is split it needs more cooling plates and increases risk of leaks, temperature deltas and coolant back pressure.

6. Thermal Interface and Mechanical Mounting

To minimize thermal interface resistance, the cold plate surface must achieve extremely tight flatness tolerances (typically < 0.05 mm). Even microscopic air gaps act as insulators, so a high-quality Thermal Interface Material (TIM) must be applied under even mounting pressure to effectively bridge the cold plate and the heat source.
- Variation in flatness => thicker TIM => poorer thermal conduction => some cells running hotter
When mounting, consider the thermal expansion/contraction, thermal conduction to the mounting points and the mechanical support of the plate and parts attached to it.