Yuichi Sei*, Koji Enoki, Seiichi Yamaguchi, Kiyoshi Saito

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review


Artificial intelligence (AI) techniques have been widely used across many fields. However, few studies have focused on the use of AI techniques for predicting heat transfer coefficients regardless of single-phase or two-phase flows. The applicability of deep neural networks [(DNNs), also known as deep learning], one of the most promising AI techniques, to horizontal-flow boiling heat transfer in mini-channels is being actively researched. The effect of surface tension in mini-channels is significant in comparison to that in conventional large tubes, and the heat transfer mechanism in the mini-channels is complicated. Thus, the accuracy of the prediction results based on existing studies is not satisfactory. Moreover, we cannot determine the uncertainty of the predicted heat transfer coefficients by using existing approaches. In this study, we propose a novel prediction mechanism, based on the combination of a DNN and Gaussian process regression, that can predict not only heat transfer coefficients with high accuracy but also the uncertainties of the predicted heat transfer coefficients. We refer to this new research field, which integrates thermal engineering and informatics, as thermoinformatics, and consider the scope of its future development.

Original languageEnglish
Pages (from-to)43-65
Number of pages23
JournalMultiphase Science and Technology
Issue number2
Publication statusPublished - 2022


  • deep learning
  • Gaussian process regression
  • heat transfer
  • thermoinformatics
  • two-phase flow boiling

ASJC Scopus subject areas

  • Modelling and Simulation
  • Condensed Matter Physics
  • Engineering(all)


Dive into the research topics of 'PREDICTION OF BOILING HEAT TRANSFER COEFFICIENTS FOR MINI-CHANNELS'. Together they form a unique fingerprint.

Cite this