Reshaping Boiling Heat Transfer Experimentation with Machine Learning and AI

The 2022 MIT Department of Nuclear Science and Engineering annual Research Expo held on April 1, 2022 showcased groundbreaking research from NSE’s various labs. The event featured research posters presented by current undergraduate and graduate students as well as signature oral presentations given by outstanding students chosen from each of the four main areas of research within the department. Graduate student, Madhumitha Ravichandran, spoke about reshaping the boiling heat transfer experimentation, leveraging Automated Science and Active Machine Learning.

[@arnabkarmakar2240]

The experimentation on a boiling flow is important to analyze its flow regimes, void fraction distribution, and pressure drop. In the practical field, the boiling flow in a channel produces sustained oscillation with a higher dimensional chaos. The high dimensional chaos is quite unpredictable in long time span. Even long term future prediction of deterministic chaos is impossible due to its dynamic nature. With an artificial intelligence like neural network tool, individualy, the dynamic behavior is difficult to predict. It only predits the static nature where it determines the flow behavior with its information at a particular time. The method shown here is effective to design a new surface and device to enhance heat transfer with safety parameter.