Data for the paper: "Deterministic drag modelling for spherical particles in Stokes regime using data-driven approaches"
Creators
-
1.
Otto-von-Guericke University Magdeburg
-
2.
University of Illinois Urbana-Champaign
Description
These are the data files for the paper:
Deterministic drag modelling for spherical particles in Stokes regime using data-driven approaches
authored by: Hani Elmestikawy, Julia Reuter, Fabien Evrard, Sanaz Mostaghim, Berend van Wachem.
The datasets encapsulate the estimated hydrodynamic forces acting on random periodic arrangements of spherical particles in Stokes regime.
Description of the data
The forces are computed using the method of periodic regularized Stokeslets. Each file includes datasets from multiple realizations at a given volume fraction, $\varepsilon_\mathrm{p}$. The files are in CSV format with the filename including the volume fraction and the total number of particles in all realizations at the given volume fraction. For example, the file data_VF_0.25_N_2775.csv includes datasets from multiple realizations at $\varepsilon_\mathrm{p}=0.25$ with total number of particles 2775. Each row in a file contains the information about an individual particle such as:
- The hydrodynamic force acting on the particle
- The relative positions of the nearest 110 neighbors (sorted from nearest to furthest).
In the datasets, the reported hydrodynamic force acting of each individual particle is normalized by the Stokes drag defined as: $$F_\text{St} = 3 \pi \mu d_\mathrm{p} u_o,$$ where
- $\mu$ is the dynamic viscosity of the fluid
- $d_\mathrm{p}$ is the particle diameter
- $u_o$ is the superficial velocity.
The description of each column in the CSV file follows:
Column |
Description |
| # | Particle id within all the datasets in this file |
instance |
simulation instance that this particle belongs to |
seed |
random seed used to generate the random arrangement |
phi |
mean particle volume fraction of the arrangement [\-] |
d_particle |
particle diameter [m] |
mu |
dynamic viscosity of the fluid [kg/(m s)] |
u |
superficial fluid velocity in the x-direction (mean flow direction) [m/s] |
v |
superficial fluid velocity in the y-direction [m/s] |
w |
superficial fluid velocity in the z-direction [m/s] |
fx/stokes |
hydrodynamic force in the x-direction normalized by $F_\text{St}$ (drag force) [\-] |
fy/stokes |
hydrodynamic force in the y-direction normalized by $F_\text{St}$ [\-] |
fz/stokes |
hydrodynamic force in the z-direction normalized by $F_\text{St}$ [\-] |
norm_rel_dist_i |
relative distance of the $i$-th neighbor normalized by $d_\mathrm{p}$ [\-] |
norm_rel_x_i |
relative x-coordinate of the $i$-th neighbor normalized by $d_\mathrm{p}$ [\-] |
norm_rel_y_i |
relative y-coordinate of the $i$-th neighbor normalized by $d_\mathrm{p}$ [\-] |
norm_rel_z_i |
relative z-coordinate of the $i$-th neighbor normalized by $d_\mathrm{p}$ [\-] |
Reading the datasets
To read the datasets, a simple python code, read_dset.py is included, where we read one of the datasets and the mean drag as an example.
Acknowledgments
The authors gratefully acknowledge the funding by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under project number 466092867 within the Priority Programme SPP 2331: Machine Learning in Chemical Engineering.
Files
zenodo-stokes-dataset.zip
Files
(54.3 MB)
| Name | Size | Download all |
|---|---|---|
|
md5:5cf714b8dd2a5e9079e9800cbabbb3b1
|
54.3 MB | Preview Download |
Additional details
Funding
- Deutsche Forschungsgemeinschaft
- Improving simulations of large-scale dense particle laden flows with machine learning: a genetic programming approach 466092867