Freja Nordsiek and Daniel P. Lathrop Collective phenomena in granular and atmospheric electrification. 2015 Nonlinear Dynamics Lab (http://complex.umd.edu) Institute for Research in Electronics and Applied Physics University of Maryland College Park, MD, USA ================================================================================ Overview ================================================================================ This repository contains data from the Granular Electrification Experiment in the University of Maryland Nonlinear Dynamics Lab. The experiment consists of a cylindrical cell with aluminum plates on the top and bottom. The cell is filled with granular particles and shaken vertically for several cycles. The vertical position of the cell and the electric potential between the top and bottom endplates of the cell are acquired. The data in this repository is from experiments in which the cylindrical cell is filled with only one type of particle. One exception uses two types of particles, pointed out below. A particle type is comprised of its material, form (spheres or powder), and size range. The acceleration timeseries of the shaking is approximately a square wave with amplitude a, meaning that the vertical position is approximately a sequence of parabolas of alternating concavity. The stroke-length of the oscillation is 10.0 cm. The shaking strength is quantified as a/g where g is the free fall acceleration due to gravity on Earth. The amount of particles is quantified by the dimensionless parameter lambda = 2 N_p d^2 / (3 D^2) where N_p is the number of particles, d is the particle diameter (or effective diameter), and D is the diameter of the cell. A manuscript reporting this data is to be submitted to a peer reviewed scientific journal and uploaded to arXiv (http://arxiv.org) with the title "Collective phenomena in granular and atmospheric electrification". This repository is mostly comprised of profiles of the cell's vertical position and the electric potential between the top and bottom plates over the shaking cycles. The oscillating motion is broken up into cycles. Each cycle begins when the cell is at its maximum vertical position and ends on the next maximum vertical position. Most of the data in this repository is broken up into the individual cycles and given as a function of phase within the oscillation cycle. These are the profiles. Each phase is approximately 1 ms apart. The actual data was acquired at 80 kHz or 120 kHz and all samples within the approximately 1 ms bins were averaged together when written to the files making up this repository, which reduces the data down to a manageable size. The repository is comprised of ten ZIP files (.zip), two CSV files (.csv), and one HDF5 file (.h5), and this TXT file (.txt). The ZIP files themselves contain a directory of the same name as the ZIP file minus the extension. The filetypes are CSV table files (.csv), HDF5 data files (.h5), JPEG image files (.jpg), PNG image files (.png), and Matlab code M-files (.m). Numerical data is stored in both CSV files (.csv) and HDF5 files (.h5). The former format is plaintext, is human readable, and can be imported by many numerical software and spreadsheet programs. The latter format is a binary format meant for scientific data interchange and is structured internally like a POSIX/Unix/Linux filesystem where Groups are equivalent to directories and Datasets are equivalent to files. The files in the root of this repository that begin with 'plaintext_' contain the plaintext CSV versions of the data, whereas the ones that begin with 'binary_' contain the binary HDF5 versions of the data. Note that the data in the plaintext CSV files is given at a lower precision. All CSV files (.csv) and HDF5 files (.h5) contain meta-data information on their contents and how to read them in the form of a header and a dedicated Dataset named '/information' respectively. All text uses DOS/Windows line endings ('\r\n'). ================================================================================ Files ================================================================================ * README.txt This file. * particle_information.csv Information on all of the particle types used in the experiment. Contains their materials, forms, sizes, and manufacturer information (including product number). * powder_characterizations.zip Particle characterization data and images for the four powders used in the experiment. The ZIP file contains four directories, one for each powder. Those directories hold the microscope images of microscope slides containing the particles (courtesy of John H. Abrahams III, NanoCenter FabLab at the University of Maryland College Park, MD, USA), images produced in the course of characterizing the particles, and the particle characterizations themselves. In the root of the ZIP file, 'particles.X' where X is 'csv' or 'h5' gives more information about the powder characterizations, the images, and the powders themselves. All data files have both a CSV version and an HDF5 version with the same names but different file extensions. * particle_arrangement_on_top.zip A picture of the particle arrangement in the cell from above after it is shaken with two particle types of the same material (69%:31% ZrO_2:SiO_2) but different sizes (200-300 micron and 500-600 micron) along with a map of the relative quantities of large particles. A quantity of lambda = 3 of each particle type were shaken together for 20000 cycles at a/g = 2.08. The picture mentioned above is 'picture.jpg'. The mapping, stored as 'mapping.png', has large values in regions dominated by large particles and small values in regions dominated by small particles. The Matlab code, 'size_mapping.m', produces the mapping and explains how it is done in detail. * binary_discharge.h5 plaintext_discharge.csv Raw timeseries of the cell vertical position and the electric potential between the plates during an electric discharge inside the cell. The timeseries is 200 ms long and is centered (t = 0) on the discharge. The cell was filled with 200-300 micron diameter 69%:31% ZrO_2:SiO_2 spheres (lambda = 16.4). The cell was shaken at a/g = 2.08. * binary_cycle_profiles.zip plaintext_cycle_profiles.zip The cycle profiles done with each particle type at a/g = 2.08 for 5000 cycles. For each particle type, there is a run at lambda = 6. Two of the particle types also have runs at higher lambdas. The ZIP file contains a directory 'runs', which then contains directories for the different lambda, which then contain the data for the runs themselves (names are derived from the particle type). The data is either a directory with CSV files 'vertical_position.csv' and 'electric_potential.csv' having the vertical position in centimeters and the electric potential between the plates in Volts respectively, or it is a single HDF5 file containing both. At the root of the ZIP file, there is a CSV file 'particles_and_files.csv' which gives more information about the ZIP file contents, the runs, and the directory and file names for each run. * binary_lambda_dependence_ps_powder_10_325.zip plaintext_lambda_dependence_ps_powder_10_325.zip The cycle profiles for polystyrene powder particles (10-325 micron effective diameter) at different lambda and a/g to see the dependence of the electric potential on those two parameters. For each quantity lambda, the particles are shaken at a/g = 2.08 for 10000 cycles (we call this the 'transient' run) before shaking the same particles at several different a/g for 1000 cycles (or 500 cycles for a/g < 0.9). The structure is similar to that of 'X_cycle_profiles.zip' above. The ZIP file contains a directory 'runs', which then contains directories for the different lambda, which then contain the data for the runs themselves. The directories for the runs themselves are named either 'transient' for the 'transient' run or after each a/g value. At the root of the ZIP file, there is a CSV file 'particles_and_files.csv' which gives more information about the ZIP file contents, the runs, and the directory and file names for each run. * binary_lambda_dependence_ps_spheres_610_990.zip plaintext_lambda_dependence_ps_spheres_610_990.zip The same as for 'X_lambda_dependence_ps_powder_10_325.zip' above except that the data is for 610-990 micron diameter polystyrene spheres and there is only one a/g done for each lambda. * binary_lambda_dependence_glass_spheres_750_1000.zip plaintext_lambda_dependence_glass_spheres_750_1000.zip The same as for 'X_lambda_dependence_ps_powder_10_325.zip' above except that the data is for 750-1000 micron diameter glass spheres and there is only one a/g done for each lambda. ================================================================================ CSV Data Structure ================================================================================ CSV - Comma Separated Value * Specification: http://www.ietf.org/rfc/rfc4180.txt * Library of Congress information: http://www.digitalpreservation.gov/formats/fdd/fdd000323.shtml CSV files represent tabular data in plain text. All CSV files in this repository start out with a header providing information/meta-data as well as the headers of each column of the table. All lines in the header start with the pound character (#) to denote them as the header, though it must be noted that CSV reading software will need to be told explicitly how many lines to skip since there is no standard convention for how to denote headers. The first line of the header indicates how many lines are in the header and takes the form '# header: X lines' where X is the number of lines taken up by the header. The tabular data follows immediately after the header lines The data is given in the table below using commas (,) as the separator between columns. Numbers are denoted like 32.3, 0.043, and 3.45e3. Strings are denoted as "hello world" in double quotes. ================================================================================ HDF5 Data Structure ================================================================================ HDF5 - Hierarchical Data Format, Version 5 * Specification: http://www.hdfgroup.org/HDF5/doc/H5.format.html * Library of Congress information: http://www.digitalpreservation.gov/formats/fdd/fdd000229.shtml * Software: https://www.hdfgroup.org/HDF5/ HDF5 files represent numerical data, text, etc. in a binary database-type file. HDF5 files are laid out like a POSIX/Unix/Linux filesystem with Groups and Datasets instead of directories and files, respectively. Textual information/meta-data for the file is always stored in the Dataset '/information' and contains similar information to the header lines in the CSV version of the file. Additional meta-data parameters (say, the particle type for the run) are contained in additional Datasets in the root Group of the folder '/'. All the numerical data is stored inside the Group '/data/'. The Datasets within it are described in '/information'. Generally, what would be stored in different files or different columns in the CSV version of the data is stored as different Datasets. For most of the HDF5 files in this repository, the Datasets in '/data/' are compressed using the built-in Deflate algorithm (no additional software or tweas are needed beyond the standard HDF5 libraries and software). Strings are stored in ASCII. Arrays are stored in C-dimension-ordering (row, column) as opposed to Fortran-dimension-ordering (column, row). ================================================================================ Licenses And Copyright ================================================================================ Copyright 2015, Freja Nordsiek and Daniel P. Lathrop If you use this data in a publication, you should cite the final journal article if it is published or the arXiv preprint if it is not published yet. It is by the aforementioned authors and bears the title "Collective phenomena in granular and atmospheric electrification". The licenses by file type/s and origins (files after unzipping the ZIP files): * CSV and HDF5 data files (.csv and .h5) CC0 (https://creativecommons.org/publicdomain/zero/1.0/) for the data inside them and CC-BY (https://creativecommons.org/licenses/by/4.0/) for the meta-data and information inside them. * Software code (Matlab M-file .m) Apache 2.0 (http://www.apache.org/licenses/LICENSE-2.0) * Images (.jpg and .png) CC-BY (https://creativecommons.org/licenses/by/4.0/), but note that the microscope images were taken by John H. Abrahams III, NanoCenter FabLab at the University of Maryland College Park, MD, USA.