Designing EV charging cables with phase change material

Task

Charging power above 150 kW means high charging currents. These heat up battery systems and charging infrastructure, causing losses: Joule heating in the copper conductors and additional re-cooling demand.

The obvious countermeasures all have drawbacks. Larger cable cross-sections add weight and material. Lower charging currents extend charging time. Active cooling, in the vehicle or in the charging station, consumes extra energy.

The study therefore looked at a passive approach. Phase change materials (PCM) buffer the heat generated during fast charging and release it later, for example while driving. The task: choose geometry and material so that a target temperature is not exceeded.

Approach

Two basic layouts were examined. In the first, the heat source sits inside and is jacketed by PCM. In the second, the PCM forms the core and the heat source surrounds it. The heat source can be, for example, a current-carrying copper conductor.

FEM models captured the transient temperature behavior. Parameters varied were inner diameter (2–15 mm), outer thickness (0.5–5 mm), charging current (100–1,000 A), and the melting temperature and melting enthalpy of the PCM (100–300 J/g). The current range covers charging power from 100 to 600 kW at battery voltages of 400 to 800 V.

A functional prototype of the second layout recorded temperatures at several points with sensors. The measurements verified the simulation and extended the data basis.

An ML model learned how geometry, material and charging current relate to the temperatures at three measuring points and three points in time. For design, inputs and outputs were swapped and the model retrained. Given boundary conditions and target temperatures, it returns the inner diameter — no classical optimization loop required.

Result

The model determined the inner diameter for a specified maximum temperature. An FEM simulation using that diameter confirmed the predicted temperature curves. According to maschinenbau 1/2024, prediction accuracy exceeded 95 %.

With a reduced conductor cross-section and a matched PCM, weight savings of up to 30 % compared with a plain copper conductor were possible for this task. Volume dropped by up to 10 %, with a similar peak temperature (ibid.).

Further variants — other temperature ranges, geometries and boundary conditions within the design space — can be assessed with the trained model in a few seconds, without new simulations.