A team of researchers in France has developed a machine-learning technique to optimize the design of thermal pump components for lower energy consumption and to achieve tremendous rotational speeds.
The proportion of new homes equipped with heat pumps in Switzerland today between 50% and 60%. These pumps draw heat energy from the surrounding environment, such as land, air, lakes and nearby rivers, and turn them into heat for buildings. The conventional heat pumps work today, and are environmentally friendly, but the field is open to improving them. Engineers can reduce the need of heat pumps to energy by 20% to 25% and reduce environmental damage, using mini turbine compressors instead of traditional pressure systems, because Turbine compressors are more efficient and ten times smaller than regular pistons. The problem is that it is difficult to assemble these small components to design heat pumps, because of their small diameter of less than 20 mm and their massive rotational speeds of more than 200,000 rpm.
A team of researchers at the Applied Mechanical Design Laboratory at the Lausanne Federal School of Applied Arts, led by Joerg Schiffmann, has developed a method that makes adding turbine compressors to heat pumps easier and faster. The researchers found simple equations to calculate the ideal dimensions of the turbine compressor in the thermal pump using one of the machine learning techniques called symbolic regression. Their recent research won the Best Paper Award at the Turbo Expo 2019 held by the American Society of Mechanical Engineers.
1500 times faster
The new researchers' method simplifies the first step in the design of turbochargers, namely the ideal size calculation and the optimal rotation speed of the desired heat pump. This is a very important step because a good initial estimate significantly reduces overall design time. Engineers typically use design schemes to measure the size of turbochargers, but the accuracy of these schemes becomes less when the equipment is smaller, and traditional designs are no longer up to date.
This is why doctoral students Violet Munier and Cyril Picard have worked on an alternative solution. They have fed machine learning algorithms with 500 000 simulations, and they have put equations to copy schemes with additional features. They are highly reliable even when turbochargers are small, as detailed as complex simulations, 1500 times. The researchers' approach also allows engineers to skip some steps in traditional design processes and pave the way for easier use of turbines in the heat pumps.
Benefits of Mini Turbo Compressors
Conventional heat pumps use pistons to compress the cooled liquid and operate using a steam pressure cycle. Pistons must be well lubricated to function properly, but oil may adhere to heat exchanger walls and impede heat transfer. Mini-air compressors, which are a few tens of millimeters in diameter without oil, run on gas bearings at the rate of hundreds of thousands of cycles per minute. The gas layers prevent friction between the components, which increases heat transfer coefficient in heat pumps by 20% to 30%.
Mini-turbochargers have been developing for several years and are almost ready. "We have been approached by many interested companies using our technology," Schiffman said. "Thanks to the researchers' work, providing the company's heat pumps with mini turbochargers will be easy."