Research Spotlight From 3D images to simulations, practical issues and software tools

From 3D images to simulations, practical issues and software tools

By Nicolas Combaret and Daniel Lichau, Visualization Sciences Group (VSG), Bordeaux, France. {nicolas.combaret, daniel.lichau}@vsg3d.com

The Visualization Sciences Group (VSG) has recently joined InterPore as new institutional member. In this research spotlight, they report on the core expertise they bring to our society.

A variety of tools are used to reliably model porous media from 3D images: image processing, segmentation, registration, quantification, preprocessing for simulation. VSG’s Avizo® Fire software is an integrated, customizable and extensible framework that supports the full image-to-simulation workflow, for prototyping and implementing routine workflows for 2D/3D image analysis and characterization.

3D image processing
Improving the quality and accurately segmenting the acquired images are key to capturing the features that are relevant for simulations:

• Images may require artifact reduction to compensate acquisition issues such as CT beam hardening or FIB-SEM shear/drift and shadowing;
• Image noise must be reduced while preserving interfaces, for instance using basic median filter or more advanced adaptive bilateral, anisotropic diffusion, or non-local mean filters;
• Many segmentation techniques can be used for optimal labelling of single or multiple phases, such as adaptive thresholding or variants of watershed algorithm.

Image-based quantification and characterization

Prior to simulations, substantial quantitative information can be directly extracted from 3D images:

• Extensive measurements of characteristics such as pore space distribution, connectivity (see fig. 1) and anisotropy, grain separation, shape factor and orientation, can be obtained directly from 3D discrete images, relying for instance on mathematical morphology algorithms;

Fig. 1: Permeable path distribution along axes (colours represent geodesic distances).

• Images using various sizes, resolutions, dimensions and modalities (e.g. micro-CT and FIB-SEM), can be registered to examine and merge material features such as porosity at different scales. Image registration also enables accurate analysis of 3D images sequences from experiments.

Modeling and numerical simulation
Physical properties can be computed from segmented 3D images by using several approaches (see fig. 2).

• Finite Element Modeling requires effective and adaptive meshing of phases from the 3D images. FEM simulation may still turn out to be computationally prohibitive for complex geometry repre-sentative of porous media.
• Pore Network Modeling is intended to be less computationally demanding. Faithful geometry approximation relies on pore space partitioning techniques, such as topology-preserving skeletonization, spatial graph extraction, and watershed (see [2]).
• Direct image-based approaches do not require demanding conversions to geometric models. Developed in collaboration with ICMCB – CNRS (D. Bernard), Avizo XLab Hydro uses a finite volume method to calculate absolute permeability in two ways: the simulation of an experiment and the calculation of the intrinsic permeability tensor – representative of property in arbitrary directions. [3] and [4] demonstrate their applicability to realistic cases.

Conclusion
Any parameter changes used for 3D image processing, filtering, reconstruction, segmentation, and representative volumes selection have an impact on the simulation results. Integrating these processes into a single framework enables more efficient and accurate evaluation of image-to-simulation workflows.

References
[1] Poromechanics Investigation at Pore-scale Using Digital Rock Physics Laboratory, S. Zhang, N. Saxena, P. Barthelemy, M. Marsh, G. Mavko, T. Mukerji, COMSOL Conference, Boston, USA. Oct 13-15, 2011
[2] Link between the microstructure of porous materials and their permeability, E. Plougonven, PhD Thesis, Bordeaux University, France. Oct, 2009
[3] Porosity and permeability analysis on nanoscale FIB-SEM 3D imaging of shale rock, S. Zhang, R. Klimentidis, P. Barthelemy, SCA Symposium 2011, Austin, USA. Sept 18-21, 2011
[4] The analysis and simulation of rock properties using FIB-SEM and virtual material studio, S. Zhang, F. Della Maestra, N. Combaret, R. Klimentidis, P. Barthelemy, R. Albou, D. Lichau, NAFEMS World Congress, Boston, USA. May 23-26, 2011