MarketPlace Use Case 2 documentation
Use case 2 consists of the simulation of viscoelastic pastes used in screen printing for solid oxide fuel cells. The entire screen-printing process is shown in Figure
Due to the complexity of the viscoelastic material, with its highly non-linear behaviour, and the high computational cost required to simulate each individual step in the process, the app focus exclusively on the levelling stage of the screen printing process.
The levelling stage is important to determine the final quality of the printing as the paste spread is an important parameter.
Paste spread is also highly dependent on paste properties, so a simulation of the levelling process is also ideal for testing different pastes and defining their optimal properties with minimal experimentation.
the Paste Levelling App takes as input files with the experimental rheometer data for different pastes and provides a measure of the paste broadening. It also provides intermediate output in the form of comparison plots between experimental and simulation rheometer data.
Check out the Usage section for further information.
Note
This project is under active development.
Contents
Description
Description of the workflow
1: Parameter estimation
Experimental rheometer data for a certain paste is taken as input for a parameter optimization script, which uses the Giesekus model equations to determine the constitutive properties required for the simulation.
2: Rheometer
The second step is a simulation of the rheometer. The constitutive properties from the parameter optimization are used in an OpenFOAM simulation with rheotools, where the rheometer is simulated and the results are compared with the experimental data for validation of the obtained constitutive properties.
3: Paste settling
Finally, the same constitutive properties are used for a full multiphase simulation of paste levelling, which is allowed to settle under gravity for the measurement of its spread.
Description of the app
The Paste Levelling Application streamlines the workflow discussed in the previous section by automating the three steps without requiring any user input between them. The application is illustrated in the figure below and is built using a series of Python modules to handle the different stages of the workflow.
The app consists internally in a series of Python modules which handle the different steps of the workflow.
Parameter optimization
The first step, parameter optimization, generates the output with the constitutive properties used in the OpenFOAM simulations of both the rheometer and levelling steps.
Rheometer simulations
The rheometer simulations are performed using a Python script that carries out a series of OpenFOAM simulations over the required range of frequencies and amplitudes to replicate the experimental rheometer data. In this step, we evaluate the ability of the Giesekus model, with parameters obtained in the previous step, to reproduce paste moduli and viscosity.
Simulation
Finally, the levelling simulation is a pre-prepared OpenFOAM case setup that requires only the appropriate constitutive properties to run. This step uses a REST API to communicate with the MarketPlace platform and the HPC app, which handles the parallel remote job.
The graphical user interface is handled via Flask, a Python-based web application framework.
Usage
When you open the app you will be presented by the following screen
To begin using the app, click on the “Paste broadening simulation” button.
Preparation
You will be presented the app interface where you can enter all the quantity you need for your case.
-Loading experimental data
The first step is to upload the experimental data in the upper part of the screen and click on Configure
-Simulation parameters
In the lower part of the screen you need to define the simulation parameters.
The parameters are described in more detail below:
Final time: This parameter sets the duration of the simulation, starting from time 0. The simulation will run until the final time is reached.
Simulation deltaT: his parameter sets the time step used in the simulation. A smaller time step will result in a more accurate simulation, but will require more computational resources.
Upward velocity: This parameter sets the upward velocity induced by the grid being lifted. The velocity is depicted in the figure below, and is a key factor in the rheometer simulation.
writeinterval: This parameter determines how often the simulation values will be saved. It must be larger than the time step.
Initial paste wide: This parameter sets the initial size of the deposed paste. This is also depicted in the figure below, and is an important input for the simulation..
Run the rheometer app: This option determines whether or not to run the parameter estimation and rheometer steps of the application. If left unchecked, the paste broadening app will run with default values.
Once everything is set, you can click on the Setup simulation button.
Running
A screen summarizing the simulation settings will appear, along with an option to run the simulation:
when you click on the Run simulation, you will see the Simulation Status change from CREATED to INPROGRESS
click on the Refresh status button, until the Simulation Status change from INPROGRESS to COMPLETED,
Results
When a simulation is completed you will be offered two option, Dowload Simulation or Postprocessing
-Download results
If you click on Dowload Simulation, you will see the screen below, which allows you to dowload the simualtion data and results.
-Postprocessing
To access the default post-processed results, simply click on the “Postprocessing” button, as shown in the image below.
Once you’ve clicked on “Postprocess results,” you will see a screen that looks similar to the one below:
The resulting plots are described in more detail below:
- Rheometer results This section shows a comparison between numerical and experimental rheometer data
Moduli: This plot compares G’ and G’’.
Viscosity: This plot compares viscosity.
- Paste simulation results: This section displays the results from simulation.
Paste broadening: This plot shows the time evolution of the normalized paste broadening, which is calculated by dividing the paste width at time t by the initial width value.
Interface comparison: This plot compares the initial interface, with the interface at the end of the simulation.