1

CT, µCT and FIB-SEM / Re: Visualization of fibers in yarns

« on: June 06, 2025, 02:29:21 PM »

Dear FabianeGommes,

Thank you for following up. Yes, indeed, we improved our pretrained neural network in FiberFind-AI to work for fibers that are not strictly aligned in x-y direction. Additionally, we extended the functionality to work for different types of fibers such as elliptical fibers. 

Best regards,

Philipp

2

CT, µCT and FIB-SEM / Re: Visualization of fibers in yarns

« on: September 06, 2023, 08:17:41 AM »

In case you have an active software maintenance you can upgrade your software version for free, otherwise please reach out to our support team such that we can send you the data individually.

3

CT, µCT and FIB-SEM / Re: Visualization of fibers in yarns

« on: September 04, 2023, 04:57:29 PM »

Dear FabianeGommes,

Thank you for your forum entry. To use the predefined neural networks for fiber identification in FiberFind some requirements for the µCT data are essential. First of all, please check if your fibers/textiles are aligned in the x/y plane and have a resolution of 6-14 voxel (for centerline network) and 6-10 voxel (for straight fibers). These requirements are based on the data the neural network has been trained on.
An exemplary data file is shipped with GeoDict and can be found in the installation folder under: Math2Market GmbH\GeoDict 2023\GrayValueDataSets\Airfilter 4µm
The provided data is a fibrous structure, but it may give you an idea of how to apply the neural network and to prepare the data accordingly.

In case there are further questions, feel free to answer to this post or reach out to our support.

Kind regards

Philipp

4

Material Modeling / Re: Dependence of porosity on particle diameter

« on: October 25, 2022, 04:17:32 PM »

Dear Alex,
Thank you for your great question. In GrainGeo-Pile we use voxelized spheres for the piling process, which can lead to small resolution effects. More problems are introduced due to the boundary conditions. To achieve comparable results the computational domain needs to be rescaled e.g. larger spheres, domain larger. Otherwise, the global packing density decreases. To generate realistic packing densities without boundary effects the top and bottom area of the structure needs to be cropped. The images below show a set of spheres, which have been piled with GrainGeo and analyzed using MatDict. The plot of the solid volume fraction highlights the influence of the top and bottom boundary effect, which can be avoided if the structure is cropped at those locations.
Regarding the definition of the stable minimum in GrainGeo-Pile we only take geometric factors into account.

 We hope this helps you get the most out of GrainGeo simulations in your future work.