Scientists in South Korea have used a computer to designed a thermoelectric generator with a design no human would have imagined sketching. I mean, the designed heat-powered generator is so unconventional that most engineers would never have thought to draw it, and yet it performs more than eight times better than traditional designs. Now, this got me asking, what if the key to better energy technology wasn’t a new material?
Precisely, these scientists are researchers at the Pohang University of Science and Technology (POSTECH), working with collaborators at the Ulsan National Institute of Science and Technology (UNIST). They have achieved these emerging technologies using advanced computational design and optimisation. The joint research team was led by Professor Jae Sung Son of the Department of Chemical Engineering at POSTECH.
The achievement, published by the authors on Nature Community and reported by TechXplore, is a developed general design framework that allows computers to autonomously identify the optimal structure of thermoelectric generators, which convert waste heat into electricity. According to the team, the device can be more than eight times as efficient as conventional designs when it comes to converting heat into electricity.
How thermoelectric generator converts heat to electricity
Heat is a byproduct of almost everything we do. Car engines generate it. Industrial plants, like steel mills and semiconductor plants, release it in huge quantities. Even our own bodies give off heat as we go about daily life. Most of that energy simply dissipates into the environment, unused.
However, scientists and engineers have been working for decades to capture some of that waste heat and turn it into electricity. Thermoelectric generators are one of the most long-regarded tools to recover that wasted heat, as they can produce power directly from a temperature difference without requiring any moving parts or additional fuel.
This has been the same principle NASA has used to power spacecraft on long missions, where reliability matters more than anything else.
Meanwhile, it's important to note that despite the implementation of thermoelectric materials to improve applications in a practical real-world sense, thermoelectric devices' performance keeps flopping and doesn't deliver results as expected. The reason is that their efficiency isn’t determined by the material alone; it also depends on how the devices are designed and structured.
Inside a thermoelectric generator, several factors are—including heat flow and the path it passes through, electrical resistance, contact losses, and external load conditions—are constantly build up and interacting with one another. However, these elements need to be carefully balanced in a highly coordinated manner for the device to perform at its full potential.
Until now, engineers traditionally tackled this complexity through human intuition and repeated experimental testing. Most thermoelectric generators, for example, are built in simple rectangular forms because they’re easy to design and manufacture. The limitation of the approach is that it relies heavily on human intuition, and it tends to explore only a narrow range of possible designs.
To overcome this limitation, the research team turned to a method known as 'computational—or topology—optimisation', a method that allows a computer to explore a vast number of possible shapes and configurations to determine the most efficient three-dimensional geometry under given conditions.
Instead of beginning with a fixed design, the computer works from the conditions the thermoelectric device will face and generates structures aimed at maximising efficiency. These conditions include factors like the thermal environment, material properties, electrical load, and even resistance at contact points between components. After the shape is generated, the computer will then continuously refine the geometry by testing and adjusting it step by step until it arrives at an optimised design suited to real-world operating conditions.
The computationally optimised resulting designs were far from conventional. I mean they didn’t look anything like the neat, boxy structures engineers are used to. Unlike the thermoelectric generators that engineers built in simple rectangular shapes because they are familiar and easy to fabricate, the computer produced a range of unconventional geometries, including the ones resembling the letter "I-shape".
Others are taking on asymmetric, hourglass-like forms—forms that would be difficult to conceive through intuition alone. At first glance, they might seem impractical or even counterintuitive. But there’s a reason behind every curve and contour.
These shapes are specifically designed by the computer to precisely control how heat and electricity move through thermoelectric devices. Once heat is guided more effectively in the devices, a larger temperature difference will be maintained to produce efficient electricity. In addition to that, the computational designs also reduce electrical resistance and minimise losses at contact points within the system.
The real-world application
The researchers fabricated and tested the computational designs using 3D-printing technology, and they were put through experimental testing to evaluate their performance. The best-performing design achieved up to 8.2 times higher efficiency compared to a conventional rectangular thermoelectric generator.
 |
| Topology optimization-based design scheme for a thermoelectric generator, along with experimental validation using 3D printing. Credit: POSTECH |
Just as importantly, the experimental result closely matched the predictions made by the computational models. That result confirms the validity of the researchers' framework and also shows that the system can be potentially scalable in the sense that wasted heat can be more effectively converted into useful electricity.
If the concept can be scaled and adapted into real-world applications, the implications are significant in the sense that industries that generate large amounts of waste heat—such as steel production, semiconductor manufacturing, and transportation—could use optimised thermoelectric systems to recover some of that lost energy.
There’s also potential for this approach to become more powerful and useful when integrated into artificial intelligence (AI). Professor Hayoung Chung clarified that "this technology can derive optimal structures directly from input conditions without human trial and error, and its range of applications and impact could expand further through integration with AI."
However, it's important to note that the computational approaches are still in their early days, and there are challenges to overcome, particularly when it comes to manufacturing and cost. But the core idea opens up new avenues for innovation.
