DEM simulations of bi-disperse beds during bedload transport

This depository contains the data of all DEM simulations used in the publication Chassagne, R., Frey, P., Maurin, R., and Chauchat, J. Mobility of bidisperse mixtures during bedload transport. Physical Review Fluids, 5(11):114307. doi:10.1103/PhysRevFluids.5.114307, as well as post processing scripts to use the data.

The simulations are located in seven folders, Monodisperse/ (mondisperse simulations where the fluid forcing is varied), N0.5/ (simulations with 0.5 layer of large particles above a small particle bed and or different fluid forcing),  N1/ (simulations with 1 layer of large particles above a small particle bed and or different fluid forcing), N2/, N3/, N4/ and sizeRatio (2 layers of large particles, fixed fluid forcing but the diameter of the underlying small particles is varied). The data of each simulations are contained in separate subfolders named after the simulation. For example, H8Sh0.45/ corresponds to a monodisperse simulation with a bedheight of 8dl (dl is the large particle diameter) and a shields number of 0.45. H10N2R2Sh0.7/ corresponds to a bidisperse simulation with a bed height of 10dl, 2 layers of large particles, a size ratio of 2 between large and small particles and a shields number of 0.7. For each simulation, the time data are saved in data.hdf5 and averaged data in average.hdf5. A GeomParam.txt file is also in each folder. It contains information of the simulation that the  post processing programm will read.

The python script used to initiate the YADE-DEM simulation  is also given for information (it contains all parameters of the simulation). The post-processing programm has been coded in python2.7 with an oriented-object procedure.  The h5py package is necessary to read the .hdf5 files. The scripts do not work in python3, but can be  very easily adapted if necessary (you only have to modify the "print" functions).  The scripts are available in ScriptsPP/ and are organized as follow. For bidisperse simualtions, a mother class in SegregationPP and two child classes SegFull (to load the full time data set) and SegMean (to load only average data). For monodisperse simualtions, a mother class in MonodispersePP and two child classes MonoFull (to load the full time data set) and MonoMean (to load only average data). Two scripts examplePP1.py and examplePP2.py are proposed and show how to manipulate theses classes and the data.

本数据集存储了发表于《Physical Review Fluids》的论文《Mobility of bidisperse mixtures during bedload transport》（作者Chassagne, R.、Frey, P.、Maurin, R.与Chauchat, J.，卷5，第11期，文章编号114307，DOI:10.1103/PhysRevFluids.5.114307）中所用的全部离散元法（Discrete Element Method, DEM）仿真数据，同时附带用于处理该数据的后处理脚本。

仿真数据存放于七个文件夹中，分别为Monodisperse/（单分散仿真，该组仿真的流体作用力参数可调）、N0.5/（在小颗粒床上方铺设有0.5层大颗粒的仿真，或流体作用力参数可调的仿真）、N1/（在小颗粒床上方铺设有1层大颗粒的仿真，或流体作用力参数可调的仿真）、N2/、N3/、N4/以及sizeRatio/（该组仿真铺设有2层大颗粒，流体作用力参数固定，但下层小颗粒的直径可变）。每个仿真的数据均存放在以仿真名称命名的独立子文件夹内。例如，H8Sh0.45/对应单分散仿真，其床高为8dl（dl为大颗粒直径），希尔兹数（Shields number）为0.45；H10N2R2Sh0.7/对应双分散仿真，其床高为10dl，铺设有2层大颗粒，大小颗粒直径比为2，希尔兹数为0.7。每个仿真文件夹内均保存有时间序列数据data.hdf5与平均数据average.hdf5，同时附带GeomParam.txt文件，该文件包含了仿真的相关参数，供后处理程序读取使用。

本文还提供了用于启动YADE-DEM仿真的Python脚本（内含仿真的全部参数）以供参考。后处理程序基于Python2.7编写，采用面向对象的编程范式。读取.hdf5文件需要依赖h5py库，且该脚本仅兼容Python2.7，若需适配Python3仅需修改print语句即可完成适配。后处理脚本存放在ScriptsPP/目录下，其结构如下：针对双分散仿真，SegregationPP目录下设有一个基类，以及两个子类SegFull（用于加载完整时间序列数据集）与SegMean（仅加载平均数据集）；针对单分散仿真，MonodispersePP目录下设有一个基类，以及两个子类MonoFull（用于加载完整时间序列数据集）与MonoMean（仅加载平均数据集）。此外还提供了两个示例脚本examplePP1.py与examplePP2.py，用于演示如何调用这些类并操作数据。