With increasing levels of electrification and autonomy in cars, thermal management systems are transitioning from reactive to proactive capabilities, deploying artificial intelligence (AI) that acts as the brain overseeing thermal requirements of multiple components as well as the passengers.

Over the last four to five decades, the evolution of automotive thermal management in cars can be seen as a long journey from simple engine cooling to a sophisticated energy management strategy. As vehicles transition from internal combustion engines (ICEs) to electric vehicles (EVs), the thermal system has shifted from being a secondary support unit to the heart of vehicle efficiency and driving range. 

For decades, thermal management was synonymous with simple heat management. This meant that primarily in early generation ICE cars, thermal management involved engine cooling and cabin cooling / heating, depending upon the regional weather conditions. This period can be roughly termed as the component-centric era, where the performance of these components was seen individually as they were decentralized. 

This was followed by the second phase of thermal technology evolution, where the industry shifted focus on integration, heat recovery and modularity as carmakers began developing dedicated energy architecture for EV-specific platforms. With integrated thermal modules becoming the mainstay, global carmakers began looking at weight reduction to optimize performance and packaging.

The second phase is expected to gradually lead us into the third phase, where artificial intelligence (AI) could act as the central brain of the car, monitoring temperatures across critical components such as the battery, electric motor, heat pumps as well as the passenger cabin.

Let’s take a look at these three pivotal phases of thermal technology evolution in vehicles in more detail:

a) The component-centric era: This was arguably the longest serving period, where thermal management systems were decentralized and components such as engine cooling loop, which involved a radiator, mechanical water pump and a thermostat, operated independently from the air conditioning (AC) loop in the cars. The OEMs focused on component and hardware system design, with performance and cost being the key drivers. Early age electric vehicles (EVs) that were developed by the global carmakers heavily borrowed from the legacy parts such as the simple AC loop. An interesting example would be the first-generation Nissan Leaf, which borrowed the AC loop from ICE cars as it was very cost-effective but weighed heavily on efficiency and performance. On the supplier front, companies such as Denso and Marelli had perfected high-volume, low-cost aluminum radiators and mechanical compressors. 

b) The part-integration, modularity era: With tightening emission norms and the overall shift towards vehicle electrification, the carmakers globally realized that thermal modules working in a car cannot be looked upon in isolation. As a result, component manufacturers evolved from build-to-print approach to joint research and development (R&D projects with the carmakers, where these partnerships experimented with component sharing and assembly of smaller parts into integrated modules to drive weight and space savings along with the costs. The integration of individual circuits such as the battery, drive unit and the cabin into a unified system resulted in the introduction of heat pump, which routes waste heat from the motors and inverter to warm the battery or cabin. In other words, this gave rise to the thermal hub where the manufacturers began using integrated thermal modules with manifolds that combined valves, pumps and sensors into a single setup. 
A relevant example of this approach would be Tesla Model Y, which deployed the octavalve setup to utilize heat across the vehicle. Tesla’s integrated heat recovery optimization allows elimination of power-hungry electric cabin heaters.

Hyundai and Kia’s E-GMP platform too is a great example where integrated heat pump system is now recognized as a benchmark for cold weather driving-range retention. Schaeffler and Mahle are among the leading suppliers of all-in-one thermal modules that car makers are increasingly using as a single, efficient unit. 

c) The AI-driven proactive era: We are currently stepping into this phase, where hardware refinement is expected to meet the software defined vehicle or SDV era. In this phase, the thermal systems are transitioning from reactive, which is responding to current temperature, to proactive (predicting future thermal requirements) needs. This era is expected to increasingly witness artificial intelligence or AI act as the brain, overseeing battery management system (BMS) and thermal controls. 

A functional example could be software processing data from GPS, weather forecasts and driving habits, to pre-cool or pre-heat components before a thermal load even occurs. A digital twin of the battery could be another meaningful example of AI taking the center-stage. In this, if AI detects a slight anomaly even in one cell heating up compared to other cells in the battery, it can adjust the thermal management to prevent degradation and ensure safety. A few interesting industry examples would include Mercedes-Benz Vision EQXX, which deploys a ‘cooling-on-demand’ aero-thermal system operated by highly advanced software; BMW and ZF jointly working to develop a predictive thermal solution that uses navigation data to prepare the battery for high-speed highway driving or DC fast charging. Bosch is integrating thermal management into its centralized vehicle computer architectures, allowing AI to balance power electronics cooling with cabin comfort in real-time. 

Recently on March 31, South Korean supplier Hanon Systems announced that it has developed a compact, multifunctional thermal management module, which is designed to integrate several components such as eCompressor, electronic expansion valve block, a combined water-cooled condenser, and internal heat exchanger, chiller AC lines and pressure and temperature sensors into a single, high-power density solution. The module, according to Hanon Systems, reduces system complexity, improves thermal performance and enhances energy utilization, contributing to extended driving range. The company said that the integrated system manages the thermal requirements of multiple vehicle subsystems through an intelligent thermal management module, which weighs only 16 kgs, dynamically regulates refrigerant flow and temperature to optimally support real time demands. Hanon’s thermal solution was first deployed in BMW’s fully electric iX3 SUV model. 

The rise of heat pumps 

While residential heat pumps have existed for decades in countries witnessing sub-zero temperatures, they were deployed in passenger cars barely 12-13 years ago. Fundamentally, a heat pump works like an air conditioner that provides heat by transferring existing thermal energy from one place to another within a car. The system deploys a refrigerant that circulates through a closed loop, charging from liquid to gas and vice versa to move heat. In summers, it pulls the heat from cabin and dumps it outside, whereas in winters, it pulls the heat from outside as well as car’s electronics to warm the passengers. 

It is known that in IC engine cars, heat is a byproduct of the engine. Since EVs had no waste heat from the engine, they originally used positive temperature coefficient or PTC heaters, which consumed significant power from the EV battery, resulting in a loss of driving range by up to 30% to 40% in freezing weather. To solve this problem, the heat pump was developed, and it is understood that the first-generation heat pumps could save 20% of that lost driving range in EVs.

It is difficult to accurately point out which carmaker pioneered the heat pump, but reports suggest that while Renault Zoe unveiled the heat pump technology at the Geneva Motor Show in March 2012, Nissan Leaf was the first production car to use the heat pump in 2013. Shortly after that, BMW i3 also offered the technology as an option in its lineup in 2013. 

The heat pumps are also vital for warming the battery to its goldilocks zone (15 degrees C to 35 degrees C) before fast charging to significantly speed up the charging time. Several carmakers have now started deploying heat pumps, which were earlier offered only in premium variants or as cold weather package option. For example, Tesla first used heat pumps in Model Y in 2020, and it became so successful that the company retrofitted it in Model 3 in 2021. 

Hyundai Motor too introduced heat pumps with their E-GMP platform in 2021 with the Ioniq 6 and Kia EV6. Volkswagen offered heat pumps in ID.3 and ID.4 as an optional ‘efficiency package’. General Motors too introduced heat pumps in its Ultium platform in 2022 with its Hummer EV and Lyriq models.

It is also important to note that heat pump technology is seen as a definitive bridge between phase 2 and phase 3 of the thermal technology evolution, as explained above. This technology is an integral part of phase 2 as it is impossible to have an integrated thermal module without a heat pump to move energy between the battery, electric motor and cabin. As a result, the heat pump helped in transforming the thermal management in a car from a collection of parts to an ecosystem. 

Moreover, in phase 3, the heat pump continues to be the hardware enabler from phase 2, but with an evolved, proactive brain. How? A typical phase 2 heat pump might turn on based on a signal from a sensor but a phase 3 heat pump, driven by AI, could turn the heat pump on 10 minutes before the passenger arrives at a charging station to charge the EV, as the central brain would know the destination.

Software: Master controller of the hardware loops

In phase 3 of the thermal management evolution, software is no longer a peripheral utility, instead it is the master controller that dictates how every joule of energy should be harvested or moved or rejected. This transition is built on three key technological pillars:

1. Route-based predictive conditioning
While traditional thermal management is reactive as it responds to a temperature spike after it happens, software-driven systems are proactive, using situational awareness to prepare the hardware in advance. The central controller integrates with GPS, real time traffic data and weather forecasts. For example, if a driver inputs a destination 50 km away that involves a steep mountain climb, followed by a DC fast charger, the software gets to work by regulating the thermal management of the battery long before the EV ascends on the hill. This helps in pre-cooling the battery and motor in anticipation of the high load climb and pre-conditions the battery to achieve its comfortable thermal range (or goldilocks zone) for the DC charger. This approach helps in terms of avoiding the thermal lag that typically slows down the charging speeds or throttles performance. 

Some interesting examples would include BMW’s predictive thermal management, Tesla’s route-based cabin preconditioning, Lexus’ predictive efficient drive for hybrid and plug-in hybrid electric models. 

Among the system suppliers, ZF’s TherMas system is a fascinating example of a software-driven, AI-based, intelligent and predictive thermal management system. First unveiled in 2023, the German supplier has been refining and improving its thermal solution, and it now claims that TherMas can extract up to one-third more driving range from the battery, even in sub-zero temperatures. The integrated module utilizes AI to accurately predict and manage the temperatures of essential EV components such as battery, electric motor, inverter and other electrical parts.

Moreover, ZF’s integrated thermal solution has compact dimensions, offering flexibility in positioning the heat pump and fluid control unit closer to the passenger cabin or near systems that have high thermal demands. This not only helps in reducing heat loss but also enhances overall thermal efficiency. This underlines hardware that is consistently refined to offer more flexibility in terms of modular and compact design, ease of installation, weight savings, among other critical parameters. ZF plans to offer TherMas in three performance classes, providing carmakers with a lot flexibility to deploy the technology across multiple EV platforms and model lineups.

Similarly, Bosch has developed its own software for predictive control of thermal system (PCTS), splitting its solution into modules that are available separately. According to the company, the software is adaptable for different vehicle classes and consists of an entry, efficiency and predictive package.

2. Digital twins and cloud-based analysis
Key thermal components now have a digital twin, which is a virtual replica stored in the cloud. The software compares real time data recorded by sensors such as temperature, flow rate, pressure, among other parameters, with the ideal performance model of the digital twin to run its own analysis. For example, if a coolant pump (in its hardware form) is drawing 5% more current than its digital twin, AI predicts a given flow rate, and the cloud-based analysis identifies this pattern as an early stage bearing wear. 

This approach helps in the form of predictive maintenance. Carmakers such as Rivian and Lucid are able to analyze thermal data from thousands of vehicles in the cloud to discover that a specific valve timing can improve efficiency by 2%. They then push that optimization back to the entire fleet via over-the-air or OTA update.

3. Neural network and AI control
This is the most cutting-edge pillar of the three, where action moves away from ‘if-then’ logic and is replaced by neural networks that learn unique thermal behavior of the vehicle and subsequent requirements. These neural networks are trained on millions of data points to understand and calculate the most efficient thermal state for the system. With the help of this complex optimization, AI can manage loops where the passenger cabin, battery, motors, and even the ADAS computer cooling are all interlinked, finding efficient thermal management a human programmer might miss. For example, if AI learns that the car owner prefers to keep 22 degrees C in the passenger cabin, but the battery needs to be at 30 degrees C for his driving style, the AI-based thermal management system can balance those conflicting needs with zero-waste energy. To sum it up, in the future, the thermal management system will be measured less by the size of the radiator and more by the millions of operations per second performed by its master controller. 

S&P Global Mobility perspective

According to Suraj Shetty, principal research analyst, thermal, S&P Global Mobility, complex coolant and refrigerant based battery thermal management is becoming standard globally – as OEMs continue to push the boundaries of battery technology and vehicle performance metrics. 

“Today Heat pump adoption is almost standard in mature EV markets like Greater China, Europe and North America, even though the configuration varies. The configuration choice has been driven mainly by refrigerant, brand engineering strategy, safety and costing,” he said.

“Integrated thermal modules are increasingly seen as a promising solution to improve system efficiency, packaging and optimizing manufacturing cost. Higher system integration also allows for more optimized system control of the thermal management system. Control strategy has evolved from early Reactive hardware & fixed controls to the current Logic Controllers & PID Regulations which offer multiple reactive optimization maps working off an array of sensors and actuators. The next step in thermal management is expected to be AI taking over the decision making in real-time, making the solutions predictive rather than reactive,” Shetty concluded.

About the author:

Amit Panday
Senior Research Analyst, S&P Global Mobility