The Higher Technical School of Engineering (ETSI) has awarded the prize to the scientific article of the ETSI quarter: October-December 2024, former Aequo to the following works:
* "A Numerical Study On Heat Transfer for Serpentine-Type Cooling Channels in A Pem Fuel Cell Stack" , Energy, Vol. 307, October 2024, pp.
13263 I​ ​Machine Learning, vol. 46, December 2024, pp. 9630-9647. DOI: 10.1109/TPAMI.2024.3422209, developed by Raúl Tapia, José Ramiro Martínez de Dios and Aníbal Ollero.

"A numerical Study on Heat Transfer for Serpentine-Type Cooling Channels in a Pem Fuel Cell Stack"

The purpose of the study is to analyze numerically the heat transfer in serpentine cooling channels in PEM type fuel battery stacks, studied the effect of the type of refrigerant, flower, input temperature, presence of thermal resistance of contact and gase diffusion layer and the bipolar plate material with CFD simulations in a cell with an active area of ​​100m2. A new correlation for the Nusselt number has been developed. The study results determine that the refrigerant flow and the thermal conductivity of the bipolar plate affects the refrigeration capacity to a greater extent. It has been concluded that for values ​​greater than 3.65% of the uniform temperature index, the differences in membrane temperatures are greater than 5 k, which could cause serious degradation problems.

The jury of the contest has highlighted the high quality of the publications presented, underlining both the scientific level of the magazines in which they were published and the relevance of the appointments received by the different works. In this context, and after applying various bibliometric indicators, the awarded work was unanimously selected for its theoretical contribution to the field of engineering.

This recognition is part of ETSI's effort to foster and make excellence in the scientific production of its research community.

 "EFFT: An Event-Based Method for the Efficient Computation of Exact Fourier Transforms"

The work introduces EFFT, the first method capable of calculating in real time and exactly the Fourier transform of an asynchronous event flow (data generated by neuromorphic cameras) without the need to convert them into conventional images. The key is a data tree that keeps the RADIX-2 decomposition of the FFT updated: only the nodes affected by each new event or event package are recalculated, reusing the rest of the results and avoiding redundant operations.
MAIN RESULTS
- Numerical accuracy: PSNR> 57DB and SSIM> 0.9.
- Speed: of the order of microsecond per event in resolution 128x128.
- Online execution in restricted hardware without loss of information.
- Open source for the research community.
IMPACT
- Allows frequency processing on event chambers in embedded systems and very low latency.
- Demonstrated in noise reduction tasks, object monitoring, visual record and scenes classification.
- It facilitates new neuromorphic vision algorithms in a wide range of applications.