Open AccessPredicting the Behaviour of a Vortex Shedding-Based Passive Mechanical Micro Heat Exchanger
Author(s) -
Francisco-Javier Granados-Ortiz,
Marina Garcia-Cardosa,
J. Ortega-Casanova
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1730/1/012098
Subject(s) - vortex shedding , microscale chemistry , reynolds number , vortex , mixing (physics) , mechanics , heat exchanger , flow (mathematics) , computer science , computational fluid dynamics , mechanical engineering , mathematics , engineering , physics , turbulence , mathematics education , quantum mechanics
In the recent years, microscale applications are gaining increasing importance. Despite their advances, the technology and resources needed to develop new designs may be a drawback for reduced scale engineering testing. To overcome this, computational methods are an efficient tool to predict how a real-life system may behave prior physical construction. The present work aims to investigate numerically effective models to predict the conditions at which a micro heat exchanger (MHE) is able to promote mixing by vortex shedding mechanics. In spite of vortex-shedding is a well-known mechanism in flow physics, it is not possible to know a priori whether a configuration (for a given geometry and flow velocity) may or may not lead to this desired vortex detachment to enhace mixing. Thus, Machine Learning methods are used for prediction, trained with finite-volume numerical simulations of different MHE devices selected based on their performance. A classification model is used to predict which configurations lead to vortex shedding. Also, a correlation regression model is developed to predict the critical Reynolds number. When the critical Reynolds number is surpassed for a given geometry, vortex shedding appears and its intensity controls the thermal mixing efficiency of the microdevice. These predictors could be useful in the search of optimal configurations by optimisation algorithms, since in the sampling process could be used to define constrains.