-
Failure modes of irregular ceramic foam under compression: Development of a new image based strut segmentation strategy
Authors:
Vinit Vijay Deshpande,
Romana Piat
Abstract:
The work investigates the failure modes of the microstructure of an irregular ceramic foam subjected to uniaxial compression loading. The foam material is manufactured using the direct foaming method and has polydispersed pores homogeneously distributed in the microstructure. The effective stress-strain curve and the macroscopic strength of the material differs significantly with the predictions o…
▽ More
The work investigates the failure modes of the microstructure of an irregular ceramic foam subjected to uniaxial compression loading. The foam material is manufactured using the direct foaming method and has polydispersed pores homogeneously distributed in the microstructure. The effective stress-strain curve and the macroscopic strength of the material differs significantly with the predictions of the Gibson-Ashby model that assumes a regular microstructure. The objective of this work is to determine the reasons. The work builds upon an image segmentation algorithm that utilizes a skeletonization method followed by geometric pruning strategies to isolate the struts in the foam microstructure. In this work, a novel pruning strategy defined by a physics-based significance measure is proposed, which identifies the struts whose failure leads to macroscopic failure of the material. The stresses in the struts are calculated by a finite element simulation, which is then utilized to determine their failure modes. This also reveals the relationship between the strut orientation and their failure modes.The energy dissipated by the individual failure modes as the loading is increased shows that there are two dominant modes which are different from the bending failure reported in the Gibson-Ashby model. These failure modes are mixed compression-tension and pure compression.
△ Less
Submitted 12 June, 2025;
originally announced June 2025.
-
Compression failure of porous ceramics: A computational study about the effect of volume fraction on damage evolution and failure
Authors:
Vinit Vijay Deshpande,
Romana Piat
Abstract:
The work describes a numerical method to study the nature of compressive failure of porous ceramics in relation to its volume fraction. The microstructure of an alumina-based foam material manufactured by mechanical stirring of a slurry is studied here. A finite element-based compressive failure simulation of a real microstructure obtained from microtomography (micro-CT) scans is conducted. A rece…
▽ More
The work describes a numerical method to study the nature of compressive failure of porous ceramics in relation to its volume fraction. The microstructure of an alumina-based foam material manufactured by mechanical stirring of a slurry is studied here. A finite element-based compressive failure simulation of a real microstructure obtained from microtomography (micro-CT) scans is conducted. A recently developed microstructure reconstruction algorithm is utilized to generate artificial microstructures statistically equivalent to the real one obtained from micro-CT. The accuracy of the reconstruction procedure is established by comparing the simulated compression behavior of the reconstructed microstructure with that of the real one along with the experimentally measured results. The effect of sample size on the simulated compression behavior is studied by computing compression stress-strain behavior for varying sizes of the reconstructed microstructures. Further, artificial microstructures of the porous ceramic with different volume fractions are reconstructed along with computing compression stress-strain behavior to establish relationship between ceramic content (volume fraction) and compressive strength of this material. The nature of the compressive failure for microstructures with different volume fractions is studied and the results are compared with the analytical models and the experimental observations available in the literature.
△ Less
Submitted 11 March, 2025;
originally announced March 2025.
-
A skeletonization based image segmentation algorithm to isolate slender regions in 3D microstructures
Authors:
Vinit Vijay Deshpande,
Romana Piat
Abstract:
The work proposes an image segmentation algorithm that isolates slender regions in three-dimensional microstructures. Characterizing slender regions in material microstructures is an extremely important aspect in material science because these regions govern the macroscopic behavior of materials for many applications like energy absorption, activation of metamaterials, stability of high temperatur…
▽ More
The work proposes an image segmentation algorithm that isolates slender regions in three-dimensional microstructures. Characterizing slender regions in material microstructures is an extremely important aspect in material science because these regions govern the macroscopic behavior of materials for many applications like energy absorption, activation of metamaterials, stability of high temperature filters, etc. This work utilizes skeletonization method to calculate centerline of the microstructure geometry followed by a novel pruning strategy based on cross-sectional area to identify slender regions in the microstructure. 3D images of such microstructures obtained from micro-CT often suffer from low image resolution resulting in high surface noise. The skeleton of such an image has many spurious skeletal branches that do not represent the actual microstructure geometry. The proposed pruning method of cross-sectional area is insensitive to surface noise and hence is a reliable method of identifying skeletal branches that represent the slender regions in the microstructure. The proposed algorithm is implemented on a test case to showcase its effectiveness. Further it is implemented on a 3D microstructure of ceramic foam to identify the slender regions present in it. It is shown that the method can be used to segment slender regions of varying dimensions and to study their geometric properties.
△ Less
Submitted 7 March, 2025;
originally announced March 2025.
-
Enhancing Canine Musculoskeletal Diagnoses: Leveraging Synthetic Image Data for Pre-Training AI-Models on Visual Documentations
Authors:
Martin Thißen,
Thi Ngoc Diep Tran,
Ben Joel Schönbein,
Ute Trapp,
Barbara Esteve Ratsch,
Beate Egner,
Romana Piat,
Elke Hergenröther
Abstract:
The examination of the musculoskeletal system in dogs is a challenging task in veterinary practice. In this work, a novel method has been developed that enables efficient documentation of a dog's condition through a visual representation. However, since the visual documentation is new, there is no existing training data. The objective of this work is therefore to mitigate the impact of data scarci…
▽ More
The examination of the musculoskeletal system in dogs is a challenging task in veterinary practice. In this work, a novel method has been developed that enables efficient documentation of a dog's condition through a visual representation. However, since the visual documentation is new, there is no existing training data. The objective of this work is therefore to mitigate the impact of data scarcity in order to develop an AI-based diagnostic support system. To this end, the potential of synthetic data that mimics realistic visual documentations of diseases for pre-training AI models is investigated. We propose a method for generating synthetic image data that mimics realistic visual documentations. Initially, a basic dataset containing three distinct classes is generated, followed by the creation of a more sophisticated dataset containing 36 different classes. Both datasets are used for the pre-training of an AI model. Subsequently, an evaluation dataset is created, consisting of 250 manually created visual documentations for five different diseases. This dataset, along with a subset containing 25 examples. The obtained results on the evaluation dataset containing 25 examples demonstrate a significant enhancement of approximately 10% in diagnosis accuracy when utilizing generated synthetic images that mimic real-world visual documentations. However, these results do not hold true for the larger evaluation dataset containing 250 examples, indicating that the advantages of using synthetic data for pre-training an AI model emerge primarily when dealing with few examples of visual documentations for a given disease. Overall, this work provides valuable insights into mitigating the limitations imposed by limited training data through the strategic use of generated synthetic data, presenting an approach applicable beyond the canine musculoskeletal assessment domain.
△ Less
Submitted 12 September, 2024;
originally announced September 2024.
