1. Committing Numerical Data to a Database (from a Simulation)

1.1. Use-Case description

Sub-System: LIGGGHTS (distributed by DCS Computing GmbH)

This use case describes how to commit data (incl. the metadata, simulation’s input and reference parameters, all the simulation output files -either heavy raw outputs or averaged outputs-, and the status/validity of the results) to a database to be later used in a workflow management tool, e.g., to calibrate DEM models.

Section

Comment

Use Case Name

Committing numerical data to database

Scope

Loading information about data and metadata and transfer to a database

Level

Top-Level data submission

Primary Actor

Researchers who wants to calibrate a DEM model, or analyse data of DEM simulations

Stakeholders and Interests

Researchers who perform DEM simulations or develop DEM models, experimentalists who conduct tests for calibration of DEM models

Preconditions

All the required input files, log files and outputs of the DEM model are available (DEM simulations are finished)

Success Guarantee

Accessing and committing data to the database is completed without error. Data is available in the database

Main Success Scenario

Meta data file is created successfully

Extensions

-

Special Requirements

none

Technology and Data Variations List

Standard input/output/log files of LIGGGHTS

Frequency of Occurrence

after every successful LIGGGHTS simulation run

Miscellaneous

-

1.2. Workflow description

In order to store any numerical data to the data base, first a metadata file should be created to store the essential information about the committed data. For this aim, critical information about each submitted numerical dataset is obtained form the logfile of the executed simulation and is written to a metadata.json file with a unique identifier (uuid). This is done by “metadatawriter” python script.

The data are mainly categorized based on their flow situations, in the database and structured as below:

Master Layout: database_CALIPER/<flow situation>/

for instance:

  • database_CALIPER/shearLE_01

  • database_CALIPER/uniComp_01/

And then are stored based on the different properties: database_CALIPER/<flow situation>/<main geometrical feature of the situation>/<particle prop 1>/<particle prop 2>/<softwareShortCut_first 7-digit uuid>

e.g. :database_CALIPER/shearLE_01/phi_0.990/fric_0.00/nu_0.00/lig_shortuuid/