## Abstract

At the beginning of the second half of the twentieth century, Proudman and Pearson (J. Fluid. Mech.,2(3), 1956, pp.237–262) suggested that the functional form of the drag coefficient (C_{D}) of a single sphere subjected to uniform fluid flow consists of a series of logarithmic and power terms of the Reynolds number (Re). In this paper, we will explore the validity of the above statement for Reynolds numbers up to 10^{6} by using a symbolic regression machine learning method. The algorithm is trained by available experimental data and data from well-known correlations from the literature for Re ranging from 0.1 to 2×10^{5}. Our results show that the functional form of C_{D} contains powers of log(Re), plus the Stokes term. The logarithmic C_{D} expressions can generalize (extrapolate) better beyond the training data than pure power series of Re and are the first in the literature to predict with acceptable accuracythe onset of the rapid decrease (drag crisis) of C_{D} at high Re, but also to follow the right behaviour towards zero Re. We also find a connection between the root of the Re-dependent terms in the C_{D} expression and the first point of laminar separation. The generalization behaviour of power-based drag coefficient equations is worse than logarithmic-based ones, especially towards the zero Re regime in which they give non-physical results. The logarithmic based C_{D} correctly describes the physics from the low Re regime to the onset of the drag crisis. Also, by applying a minor modification in the logarithmic based equations, we can predict the drag coefficient of an oblate spheroid in the high Re regime.

Original language | English |
---|---|

Article number | 118195 |

Number of pages | 21 |

Journal | Chemical Engineering Science |

Volume | 265 |

DOIs | |

Publication status | Published - 2023 |

## Keywords

- Drag coefficient
- Machine learning
- Matched asymptotic expansions
- Multi-phase flows
- sphere