Identification of the local mechanical properties of an inorganic fiber-based material using X-ray computed micro tomography and discrete elements method
Segmentation d images de documents anciens par approche texture - Mo…Cédric Mouats
Ce travail présente une méthode de segmentation floue des images de documents anciens. Cette méthode permet la séparation des zones de texte et de dessins d’images de documents imprimés datant de la Renaissance. L’approche proposée consiste à définir des bancs de filtres de Gabor capables de localiser les zones de textes et de dessin séparément à l’aide d’un processus de classification floue des résultats de filtrage. Une simple fusion des résultats des bancs de filtres fournit une version segmentée de l’image de document ancien en question.
3D Characterisation of Void Distribution in Resin Film Infused CompositesFabien Léonard
X-ray computed tomography was used to characterize the pore distribution in two composite panels manufactured by resin film infusion - one with a low viscosity resin (Panel L) and one with a high viscosity resin modified to increase toughness (Panel H). The analysis found:
- Both panels had voids mainly located around the yarns, with a sharp peak in the void distribution around 0.12mm from yarns.
- Panel L had a homogeneous void distribution, while Panel H had higher void concentrations closer to the center.
- 50% of voids in Panel L were within 2.5mm of the middle, versus 0.5mm for Panel H. 90% of voids were within 0.
Fault identification in transformer windingeSAT Journals
This document discusses fault identification in transformer windings during impulse testing. It presents an analysis of transformer winding current waveforms using both time-domain and frequency-domain (Fast Fourier Transform) methods to classify insulation failures. Simulation results are shown for a range of distribution transformer models with different fault types created at various locations. The proposed approach is applied to an analog model of a 12kVA single-phase transformer, with faults detected by analyzing the neutral currents under applied low voltage impulses.
The document describes the manufacturing process of 500 MW turbo generator stator winding bars at BHEL Haridwar. It discusses the 8 blocks involved in manufacturing including the electrical machine block, fabrication block, and coil & insulation manufacturing block. Within the coil & insulation block, it details the various sections for assembling, winding stator bars, impregnation, and testing the completed bars including tan delta and high voltage tests before the bars are installed in the generator. It provides key technical specifications for the 500 MW turbo generators.
This is my Summer Training Report in B.H.E.L., Haridwar on General Awareness Of Steam Turbine in Turbine Block-III in the period 17/06/13 to 31/07/13.
I would like to thank My Mentor Mr. Alok Shukla Sir, My Guide Asst. Prof. Mr. Anuj Dixit Sir. Without their guidance and efforts it was impossible to achieve.
Segmentation d images de documents anciens par approche texture - Mo…Cédric Mouats
Ce travail présente une méthode de segmentation floue des images de documents anciens. Cette méthode permet la séparation des zones de texte et de dessins d’images de documents imprimés datant de la Renaissance. L’approche proposée consiste à définir des bancs de filtres de Gabor capables de localiser les zones de textes et de dessin séparément à l’aide d’un processus de classification floue des résultats de filtrage. Une simple fusion des résultats des bancs de filtres fournit une version segmentée de l’image de document ancien en question.
3D Characterisation of Void Distribution in Resin Film Infused CompositesFabien Léonard
X-ray computed tomography was used to characterize the pore distribution in two composite panels manufactured by resin film infusion - one with a low viscosity resin (Panel L) and one with a high viscosity resin modified to increase toughness (Panel H). The analysis found:
- Both panels had voids mainly located around the yarns, with a sharp peak in the void distribution around 0.12mm from yarns.
- Panel L had a homogeneous void distribution, while Panel H had higher void concentrations closer to the center.
- 50% of voids in Panel L were within 2.5mm of the middle, versus 0.5mm for Panel H. 90% of voids were within 0.
Fault identification in transformer windingeSAT Journals
This document discusses fault identification in transformer windings during impulse testing. It presents an analysis of transformer winding current waveforms using both time-domain and frequency-domain (Fast Fourier Transform) methods to classify insulation failures. Simulation results are shown for a range of distribution transformer models with different fault types created at various locations. The proposed approach is applied to an analog model of a 12kVA single-phase transformer, with faults detected by analyzing the neutral currents under applied low voltage impulses.
The document describes the manufacturing process of 500 MW turbo generator stator winding bars at BHEL Haridwar. It discusses the 8 blocks involved in manufacturing including the electrical machine block, fabrication block, and coil & insulation manufacturing block. Within the coil & insulation block, it details the various sections for assembling, winding stator bars, impregnation, and testing the completed bars including tan delta and high voltage tests before the bars are installed in the generator. It provides key technical specifications for the 500 MW turbo generators.
This is my Summer Training Report in B.H.E.L., Haridwar on General Awareness Of Steam Turbine in Turbine Block-III in the period 17/06/13 to 31/07/13.
I would like to thank My Mentor Mr. Alok Shukla Sir, My Guide Asst. Prof. Mr. Anuj Dixit Sir. Without their guidance and efforts it was impossible to achieve.
1) The document discusses using autocorrelation and rose diagrams to analyze the structure and anisotropy of polymer foams from X-ray microtomography data.
2) Autocorrelation provides an efficient way to measure characteristic lengths and identify patterns in noisy 3D tomography data without needing segmentation.
3) Rose diagrams map the autocorrelation in all directions to provide a global view of preferential orientation of features and anisotropy.
The document discusses using X-ray micro-computed tomography (μCT) to study the thermomechanical properties of thermostructural composites. It describes how μCT is used to generate 3D images of composite microstructures, from which fiber orientations and material densities are measured. Finite element models are then enriched with this microstructural data and used to calculate the composites' thermal dilation and mechanical behavior. The approach is demonstrated for carbon/carbon and SiC/SiC composites. Capturing real material microstructures enables accurate thermomechanical property predictions.
The document summarizes research on characterizing the microstructure evolution of cast AlMgSi alloys using synchrotron tomography. Key findings include: (1) Synchrotron tomography was used to investigate microstructure evolution during solidification and heat treatment, (2) Primary α-Al dendrites and eutectic α-Al/Mg2Si formed with a highly interconnected seaweed-like morphology, (3) During heat treatment, the eutectic phases spheroidized and the contiguity between Mg2Si and Si remained.
This document discusses defects in cast aluminum alloys and methods for their non-destructive evaluation. It presents an outline on pores in cast aluminum components, the influence of defects on fatigue life, developing parametric models using computer tomography, and correlating finite element models with experiments. The goal is to better understand crack-initiating defects in cast aluminum alloys through non-destructive testing and modeling.
This document describes building a pore network model from 3D images of a pore space to precisely predict permeability. Key steps include:
1) Skeletonizing the 3D image to extract the pore network topology.
2) Partitioning the pore space to identify individual pores and throats.
3) Constructing the pore network model (PNM) graph from the skeleton and partitioning.
4) Computing local resistances within the PNM to predict permeability and comparing with direct numerical modeling.
This document discusses early embryo development in Arabidopsis thaliana. It aims to quantitatively describe key events related to geometry and mechanics during this process. Specifically, it seeks to:
1. Determine the sequence of developmental events by reconstructing 3D cell dynamics over time. This involves cell segmentation, lineage tracking, and growth modeling.
2. Understand why these events occur in this particular order by exploring potential biophysical and mechanical influences, such as cell curvature and distorsions during shaping.
The findings could provide insights into a few fundamental rules governing early embryogenesis through simulations incorporating geometry from 3D reconstructions.
The document discusses climate modeling and simulations performed at the German Climate Computing Center (DKRZ). It provides an overview of DKRZ's high performance computing capabilities and describes the components and coupling of Earth system models. It also summarizes simulations contributed to the CMIP5/IPCC AR5 project, including temperature and sea ice projections under different scenarios. Visualization of simulation output is discussed along with the Avizo Green software developed in collaboration with DKRZ.
The document describes the Visualization Laboratory at King Abdullah University of Science and Technology. It provides an overview of the laboratory's core facilities and capabilities for visualizing large datasets from various research centers at the university, including biology, imaging, materials science, chemical science, mechanical engineering, and more. The laboratory houses advanced visualization facilities like a 24 projector dome and 96 GPU computing cluster to enable interactive exploration and analysis of petabyte-scale scientific data.
The document discusses using CT scanning and 3D shape analysis to classify carbonate rock pores. It introduces CT scanning workflow and principles, showing how it provides 3D quantitative and qualitative pore structure data. Pore shapes are mathematically described using ellipsoid fitting of principal moments of inertia to calculate dimensions L, I, and S. Shape classes are then defined based on ratios of these dimensions. The analysis aims to better characterize carbonate reservoir heterogeneity at different scales.
The document summarizes the evaluation of rock properties and structures at the micrometer scale using sub-micrometer X-ray computed tomography. It introduces a nanofocus X-ray tube capable of less than 800 nm spot size and a nanoCT system used to scan geological rock samples with resolutions under 1 micrometer. Example scans of Bentheimer sandstone show individual mineral phases like quartz, clay, and feldspar. Scans of pyroclastic rocks resolve pores and fractures less than 2 micrometers thick. The high resolution CT allows analyzing pore structure, surface area, and simulating fluid flow at the microscale for understanding rock physics.
The document describes methods for improving X-ray contrast of phases in porous rock samples using micro-CT imaging and image processing. Heavy metal ions are used to enhance the contrast of ice and clays. A Bruker micro-CT scanner with cooling stage is used to image ice-saturated rock samples. Image segmentation software Avizo Fire is applied to separate phases and calculate properties like porosity, permeability, and pore size distribution from the 3D digital models.
The document describes experiments using an environmental scanning electron microscope to generate 3D reconstructions of membrane structures from 2D image slices. Membrane samples were embedded in resin and ultramicrotomed into thin sections for imaging. Image stacks from multiple samples were assembled into 3D models and analyzed to calculate membrane pore characteristics and water flux measurements, validating the 3D reconstruction method. The results provide a quantitative view of membrane nanostructure-property relationships not possible with conventional techniques.
This document describes the capabilities of the FIB-Nanotomography facility at the Centre Interdisciplinaire de Microscopie Electronique (CIME) at the École Polytechnique Fédérale de Lausanne (EPFL). The facility contains a Zeiss NVision 40 dual beam FIB/SEM that allows for the automated acquisition of large 3D volumes with voxel sizes down to 5-10nm. Examples are given of its applications in materials science, including the reconstruction of microstructures in superconducting cables and solder joints. The facility is also used for life science applications such as serial sectioning of brain tissue to reconstruct neuronal structures at nanoscale resolution. Automated segmentation techniques are applied to
This document discusses using X-ray computed tomography to characterize the internal structure of asphalt. Key points:
- X-ray CT is used to capture images of asphalt with micrometer-scale resolution, allowing segmentation of stones, binder, and air voids.
- Parameters describing the stone size distribution, shape, contacts, and air void structure are obtained from the segmented images.
- Finite element modeling incorporates the structural information to quantify stresses and strains in each phase under loading and temperature changes.
- Understanding the internal structure-property relationships enables improved mixture design and prediction of field performance.
1) The document discusses using autocorrelation and rose diagrams to analyze the structure and anisotropy of polymer foams from X-ray microtomography data.
2) Autocorrelation provides an efficient way to measure characteristic lengths and identify patterns in noisy 3D tomography data without needing segmentation.
3) Rose diagrams map the autocorrelation in all directions to provide a global view of preferential orientation of features and anisotropy.
The document discusses using X-ray micro-computed tomography (μCT) to study the thermomechanical properties of thermostructural composites. It describes how μCT is used to generate 3D images of composite microstructures, from which fiber orientations and material densities are measured. Finite element models are then enriched with this microstructural data and used to calculate the composites' thermal dilation and mechanical behavior. The approach is demonstrated for carbon/carbon and SiC/SiC composites. Capturing real material microstructures enables accurate thermomechanical property predictions.
The document summarizes research on characterizing the microstructure evolution of cast AlMgSi alloys using synchrotron tomography. Key findings include: (1) Synchrotron tomography was used to investigate microstructure evolution during solidification and heat treatment, (2) Primary α-Al dendrites and eutectic α-Al/Mg2Si formed with a highly interconnected seaweed-like morphology, (3) During heat treatment, the eutectic phases spheroidized and the contiguity between Mg2Si and Si remained.
This document discusses defects in cast aluminum alloys and methods for their non-destructive evaluation. It presents an outline on pores in cast aluminum components, the influence of defects on fatigue life, developing parametric models using computer tomography, and correlating finite element models with experiments. The goal is to better understand crack-initiating defects in cast aluminum alloys through non-destructive testing and modeling.
This document describes building a pore network model from 3D images of a pore space to precisely predict permeability. Key steps include:
1) Skeletonizing the 3D image to extract the pore network topology.
2) Partitioning the pore space to identify individual pores and throats.
3) Constructing the pore network model (PNM) graph from the skeleton and partitioning.
4) Computing local resistances within the PNM to predict permeability and comparing with direct numerical modeling.
This document discusses early embryo development in Arabidopsis thaliana. It aims to quantitatively describe key events related to geometry and mechanics during this process. Specifically, it seeks to:
1. Determine the sequence of developmental events by reconstructing 3D cell dynamics over time. This involves cell segmentation, lineage tracking, and growth modeling.
2. Understand why these events occur in this particular order by exploring potential biophysical and mechanical influences, such as cell curvature and distorsions during shaping.
The findings could provide insights into a few fundamental rules governing early embryogenesis through simulations incorporating geometry from 3D reconstructions.
The document discusses climate modeling and simulations performed at the German Climate Computing Center (DKRZ). It provides an overview of DKRZ's high performance computing capabilities and describes the components and coupling of Earth system models. It also summarizes simulations contributed to the CMIP5/IPCC AR5 project, including temperature and sea ice projections under different scenarios. Visualization of simulation output is discussed along with the Avizo Green software developed in collaboration with DKRZ.
The document describes the Visualization Laboratory at King Abdullah University of Science and Technology. It provides an overview of the laboratory's core facilities and capabilities for visualizing large datasets from various research centers at the university, including biology, imaging, materials science, chemical science, mechanical engineering, and more. The laboratory houses advanced visualization facilities like a 24 projector dome and 96 GPU computing cluster to enable interactive exploration and analysis of petabyte-scale scientific data.
The document discusses using CT scanning and 3D shape analysis to classify carbonate rock pores. It introduces CT scanning workflow and principles, showing how it provides 3D quantitative and qualitative pore structure data. Pore shapes are mathematically described using ellipsoid fitting of principal moments of inertia to calculate dimensions L, I, and S. Shape classes are then defined based on ratios of these dimensions. The analysis aims to better characterize carbonate reservoir heterogeneity at different scales.
The document summarizes the evaluation of rock properties and structures at the micrometer scale using sub-micrometer X-ray computed tomography. It introduces a nanofocus X-ray tube capable of less than 800 nm spot size and a nanoCT system used to scan geological rock samples with resolutions under 1 micrometer. Example scans of Bentheimer sandstone show individual mineral phases like quartz, clay, and feldspar. Scans of pyroclastic rocks resolve pores and fractures less than 2 micrometers thick. The high resolution CT allows analyzing pore structure, surface area, and simulating fluid flow at the microscale for understanding rock physics.
The document describes methods for improving X-ray contrast of phases in porous rock samples using micro-CT imaging and image processing. Heavy metal ions are used to enhance the contrast of ice and clays. A Bruker micro-CT scanner with cooling stage is used to image ice-saturated rock samples. Image segmentation software Avizo Fire is applied to separate phases and calculate properties like porosity, permeability, and pore size distribution from the 3D digital models.
The document describes experiments using an environmental scanning electron microscope to generate 3D reconstructions of membrane structures from 2D image slices. Membrane samples were embedded in resin and ultramicrotomed into thin sections for imaging. Image stacks from multiple samples were assembled into 3D models and analyzed to calculate membrane pore characteristics and water flux measurements, validating the 3D reconstruction method. The results provide a quantitative view of membrane nanostructure-property relationships not possible with conventional techniques.
This document describes the capabilities of the FIB-Nanotomography facility at the Centre Interdisciplinaire de Microscopie Electronique (CIME) at the École Polytechnique Fédérale de Lausanne (EPFL). The facility contains a Zeiss NVision 40 dual beam FIB/SEM that allows for the automated acquisition of large 3D volumes with voxel sizes down to 5-10nm. Examples are given of its applications in materials science, including the reconstruction of microstructures in superconducting cables and solder joints. The facility is also used for life science applications such as serial sectioning of brain tissue to reconstruct neuronal structures at nanoscale resolution. Automated segmentation techniques are applied to
This document discusses using X-ray computed tomography to characterize the internal structure of asphalt. Key points:
- X-ray CT is used to capture images of asphalt with micrometer-scale resolution, allowing segmentation of stones, binder, and air voids.
- Parameters describing the stone size distribution, shape, contacts, and air void structure are obtained from the segmented images.
- Finite element modeling incorporates the structural information to quantify stresses and strains in each phase under loading and temperature changes.
- Understanding the internal structure-property relationships enables improved mixture design and prediction of field performance.
Asphalt internal structure characterization with X-Ray computed tomography
Identification of the local mechanical properties of an inorganic fiber-based material using X-ray computed micro tomography and discrete elements method
1. Identification of the local
mechanical properties of an
inorganic fiber‐based material using
X‐ray computed micro tomography
and distinct elements method
Hauss G., Bernard D., Dedecker F., Couprie M., Meulenyzer S., Pourcel F.
2. Outline
• Fiber‐based material
• Experimental settings
• Image processing
• Fibers labelling and contacts extraction
• Distinct Element Modeling (PFC3D)
• Numerical results
• Conclusions and perspectives
2
7. Experimental settings
• Experimental settings
Z = 170 mm
ZD = 340 mm Source
pxSize = 25 µm Detector
N = 2500
t = 1 s
U = 100 kV Spinning
I = 240 µA mandrel
7
9. Image processing
• Global procedure
Data
reconstruction
Median filter Grey level
using a Beam Thresholding
3*3*3 normalization
Hardenning
correction
9
20. Fibers labeling and contacts extraction
• Fibers labeling Image data:
‐Voxel size
‐Fiber size (Vox.)
‐Fiber radius (Vox.)
SKELETONIZATION +
IDENTIFICATION
Skeleton data
‐Fiber ID
‐Upper Fiber ID
‐Spline equation (3rd order)
[1] John Chaussard, Michel Couprie, and Hugues Talbot. A discrete lambda‐medial axis. In 15th Discrete Geometry for Computer Imagery (DGCI’09), Lecture Notes
in Computer Science, pages 1–12, 2009. To appear. 20
[2] F. Chazal and A. Lieutier. The lambda medial axis. Graphical Models, 67(4) :304–331, 2005.
21. Fibers labeling and contacts extraction
• Contacts extraction
Contacts:
Surface area of intersection of 2 labelised fibers
PROJECTION Contact description
‐ Contact ID
‐Fiber 1 ID in contact + length of
contact (S1‐S2)
‐Fiber 2 ID in contact + length of
contact (S’1‐S’2)
21
23. Distinct Element Modeling
• Contacts definitions
Intra‐fiber (2 successive particles) and detected1 inter‐fiber contacts modeling
contacts bonded !
Parallel bond model
contact
kn
k intra
* ( pb _ rad * R part ) 2
n
1 2
PB k contact kn * kn
1
kn kn
n 2
pb _ rad 0 .25
Cracks (inter‐ and intra‐fiber) and non detected2 inter‐fiber contacts modeling
Incremental linear law
Fn k n * ( un u0 ) , si un u0 u0 : Overlap t 0
Fn 0 , si un u0 un : Overlap t 0 1
1 Inter‐fibers contacts detected on real image
2 Inter‐fibers contacts non detected on real image 23
24. Numerical results
• Assumptions and parameters to calibrate
Modeling hypothesis
Fibers deformation rather than inter‐fibers contacts deformation (experimental
observations)
K intra‐fibers < K inter‐fibers
Same stiffness before and after fracture
K inter‐fibers after fracture = K intra fibers before fracture
Parameters to calibrate
Kpart K intra‐fiber
Rm intra : Intra‐fibers strength
Rm inter : Inter‐fibers strength
24
25. Numerical results
• Particle stiffness global fitting procedure
(infinite strengths elastic behavior)
Kpart = 160 kN/m
E = 56 MPa
Strength (kN)
E = 56 MPa
Strength (kN)
F = 1130 N
Vertical displacement (mm) Vertical displacement (mm)
Experimental compression test Numerical compression test
25
26. Numerical results
• Ultimate strength fitting procedure
Strength vs. displacement curve
Rm intra = 10 Mpa
Rm inter = 50 Mpa
Stregnth (kN)
Strength (N)
Fmax = 1100 N
Displacement (mm)
Experimental compression tests Displacement (mm)
Numerical compression test
26
27. Numerical results
• Rm influences on strength‐displacement curves
ductile failure
Compressive strength (kN)
Compressive strength (kN)
Compressive strength (kN)
brittle failure brittle failure
Vertical displacement (mm) Vertical displacement (mm) Vertical displacement (mm)
Rm intra = 40 MPa Rm intra = 80 MPa Rm intra = 40 MPa
Rm inter = 40 MPa Rm inter = 40 MPa Rm inter = 80 MPa
Fmax = 1100 N Fmax = 1100 N Fmax = 2200 N
27
28. Numerical results
• Modeling of sample deformation
Simulated fibers displacements Comparison with experimental data
28
30. Perspectives
• Local properties consideration
– Identification of local mechanical properties around
favorable cracks sites
– Local variability included to model:
• Volume: Stochastic distribution of defects along fiber
• Punctual: Singular defects localization
Fiber porosity Punctual defect
30
32. Conclusions
• Fitting procedures using real image data set:
– Excellent modeling of the initial sample
(geometry)
– Calibration of the micro mechanical parameters
(considering few assumptions)
• Good modeling of the macroscopic
mechanical behavior
• In the real life, high local variability of micro
parameters (stiffness and strength)
32
35. Question ITASCA
• Pour les contacts intra‐fibre:
La loi de répulsion par contact élastique linéaire est
elle toujours active ?
Si tu parles du fait que 2 particules qui se chevauchent, doivent en théorie
s’écarter, alors non la loi de répulsion est désactivée pour éviter ce
phénomène. On base le comportement sur des lois élastiques purement
incrémentales.
La loi parallel bond est elle constante lors de la
déformation des fibres ?
Les propriétés locales des PB sont constantes tout au long du calcul (quelque
soit la déformation). Elles dépendent par contre du type de contact inter
ou intra‐fibre.
36. Question ITASCA
• Pour les contacts inter‐fibre:
Comment est définie la loi parallel bond par rapport aux zones de contact
observées (redistribution de l’aire de contact) ?
Je ne suis pas certain de bien comprendre la question. Ce qui est sur c’est que nous n’avons pas
pris en compte l’aire du contact, mais simplement la coordonnée linéique de contact le long
des 2 fibres
Comment est définie la zone de contact ?
Chaque fibre est représentée par son équation paramétrique. La zone de contact d’une fibre avec
ses voisines est définie sur un intervalle [T1;T2] de son équation paramètrique
Conservation du volume de recouvrement entre les deux fibres qui impose un
intervalle de contact ajusté.
Ou conservation de l’abscisse curviligne du contact qui impose un nouveau
volume de contact ?
On conserve l’abscisse curviligne du contact entre les 2 fibres. Aucune information sur le volume
du contact n’a été intégrée !
37. Question ITASCA
• Interprétation des résultats:
Les rigidités locales sont elles utilisées pour
remonter aux données matériau ?
Le calage des rigidités locales a permis de reproduire le comportement macroscopique du
matériau (sous essai de compression). Par contre, il n’a pas été possible à ce stade de
considérer des propriétés locales différentes pour reproduire les déplacements locaux
des fibres
Rigidité particule Module d’Young du matériau.
La conclusion est un peu rapide, car ici la rigidité des particules n’intervient que lorsque les
liaisons PB se cachent (rupture d’une fibre ou d’un contact inter‐fibres). Avant cela, le
module d’Young du matériau dépend directement de la rigidité des PB inter et intra
fibres.
43. Question ITASCA
• Pour les contacts inter‐fibre:
Comment est définie la loi parallel bond par rapport
aux zones de contact observées (redistribution de
l’aire de contact) ?
Comment est définie la zone de contact ?
Conservation du volume de recouvrement entre les
deux fibres qui impose un intervalle de contact
ajusté.
Ou conservation de l’abscisse curviligne du contact
qui impose un nouveau volume de contact ?