Summary
Heat exchangers (HXs) are at the heart of many energy systems, one example being engine cooling in the aviation industry. Advanced design of HXs is urgent since aircraft systems are becoming smaller and need to become more efficient. Through topology optimisation (TO) and additive manufacturing (AM), custom compact HXs will be designed to cool the engines of tomorrow. Although TO of thermofluidic problems has recently undergone tremendous development, the technology is mostly limited to academic problems, since existing approaches are restricted to macroscopic design with extreme computational cost being prohibitive for industrial applications.
The objective of ADeHEx is therefore to propose an integrated design methodology for multi-scale 3D fluid-to-fluid HXs using machine learning-based de-homogenization. Specifically, I will a) construct a homogenized thermohydraulic computational model; b) use convolutional neural networks to recover detailed micro-channel design from homogenization-based TO, reducing computational time by at least two orders of magnitude. Implemented at University of Southern Denmark and at the secondment, Brown University, a two-way transfer of knowledge is guaranteed through my expertise in level-set-based TO and the expertise of the supervisors in complex multiphysics modelling and physics-based machine learning. ADeHEx will a) consolidate my academic excellence and professional maturity through new skills and competences in machine learning, homogenized thermohydraulic models, teaching, supervision, project management, dissemination, industrial engagement, networking; b) harness the complementary expertise of two strong multidisciplinary teams, pushing research to the forefront of design engineering and computer science and informatics; c) revolutionize the design process of HXs and multiphysics system optimisation, while benefiting EU companies in manufacturing, aviation, etc., and raising European economic competitiveness.
The objective of ADeHEx is therefore to propose an integrated design methodology for multi-scale 3D fluid-to-fluid HXs using machine learning-based de-homogenization. Specifically, I will a) construct a homogenized thermohydraulic computational model; b) use convolutional neural networks to recover detailed micro-channel design from homogenization-based TO, reducing computational time by at least two orders of magnitude. Implemented at University of Southern Denmark and at the secondment, Brown University, a two-way transfer of knowledge is guaranteed through my expertise in level-set-based TO and the expertise of the supervisors in complex multiphysics modelling and physics-based machine learning. ADeHEx will a) consolidate my academic excellence and professional maturity through new skills and competences in machine learning, homogenized thermohydraulic models, teaching, supervision, project management, dissemination, industrial engagement, networking; b) harness the complementary expertise of two strong multidisciplinary teams, pushing research to the forefront of design engineering and computer science and informatics; c) revolutionize the design process of HXs and multiphysics system optimisation, while benefiting EU companies in manufacturing, aviation, etc., and raising European economic competitiveness.
Unfold all
/
Fold all
More information & hyperlinks
| Web resources: | https://cordis.europa.eu/project/id/101106842 |
| Start date: | 01-10-2023 |
| End date: | 30-09-2025 |
| Total budget - Public funding: | - 214 934,00 Euro |
Cordis data
Original description
Heat exchangers (HXs) are at the heart of many energy systems, one example being engine cooling in the aviation industry. Advanced design of HXs is urgent since aircraft systems are becoming smaller and need to become more efficient. Through topology optimisation (TO) and additive manufacturing (AM), custom compact HXs will be designed to cool the engines of tomorrow. Although TO of thermofluidic problems has recently undergone tremendous development, the technology is mostly limited to academic problems, since existing approaches are restricted to macroscopic design with extreme computational cost being prohibitive for industrial applications.The objective of ADeHEx is therefore to propose an integrated design methodology for multi-scale 3D fluid-to-fluid HXs using machine learning-based de-homogenization. Specifically, I will a) construct a homogenized thermohydraulic computational model; b) use convolutional neural networks to recover detailed micro-channel design from homogenization-based TO, reducing computational time by at least two orders of magnitude. Implemented at University of Southern Denmark and at the secondment, Brown University, a two-way transfer of knowledge is guaranteed through my expertise in level-set-based TO and the expertise of the supervisors in complex multiphysics modelling and physics-based machine learning. ADeHEx will a) consolidate my academic excellence and professional maturity through new skills and competences in machine learning, homogenized thermohydraulic models, teaching, supervision, project management, dissemination, industrial engagement, networking; b) harness the complementary expertise of two strong multidisciplinary teams, pushing research to the forefront of design engineering and computer science and informatics; c) revolutionize the design process of HXs and multiphysics system optimisation, while benefiting EU companies in manufacturing, aviation, etc., and raising European economic competitiveness.
Status
SIGNEDCall topic
HORIZON-MSCA-2022-PF-01-01Update Date
31-07-2023
Images
No images available.
Geographical location(s)
