2015 USNC-URSI Radio Science Meeting (Joint with AP-S Symposium), 2015
Electromagnetic scattering from Carbon Nanotubes (CNT) has received wide interest in the past dec... more Electromagnetic scattering from Carbon Nanotubes (CNT) has received wide interest in the past decade. Many different CNT configurations have been computationally investigated such as single CNTs, infinite planar arrays of CNTs, finite arrays with simple distributions, and bundles of CNTs. In all of the previously reported configurations, the CNTs were perfectly straight and they were arranged in a uniform distribution. However, in commercial CNT composites the CNTs typically exhibit highly complex shapes and distributions. The goal of this work is to simulate and characterize the electromagnetic scattering from multiple CNTs with realistic shapes and distributions that resemble those found in commercial composites.
ABSTRACT A numerical task of current interest is to compute the effective elastic properties of a... more ABSTRACT A numerical task of current interest is to compute the effective elastic properties of a random composite material by operating on a 3D digital image of its microstructure obtained via X-ray computed tomography (CT). The 3-D image is usually sub-sampled since an X-ray CT image is typically of order 10003 voxels or larger, which is considered to be a very large finite element problem. Two main questions for the validity of any such study are then: can the sub-sample size be made sufficiently large to capture enough of the important details of the random microstructure so that the computed moduli can be thought of as accurate, and what boundary conditions should be chosen for these sub-samples? This paper contributes to the answer of both questions by studying a simulated X-ray CT cylindrical microstructure with three phases, cut from a random model system with known elastic properties. A new hybrid numerical method is introduced, which makes use of finite element solutions coupled with exact solutions for elastic moduli of square arrays of parallel cylindrical fibers. The new method allows, in principle, all of the microstructural data to be used when the X-ray CT image is in the form of a cylinder, which is often the case. Appendix A describes a similar algorithm for spherical sub-samples, which may be of use when examining the mechanical properties of particles. Cubic sub-samples are also taken from this simulated X-ray CT structure to investigate the effect of two different kinds of boundary conditions: forced periodic and fixed displacements. It is found that using forced periodic displacements on the non-geometrically periodic cubic sub-samples always gave more accurate results than using fixed displacements, although with about the same precision. The larger the cubic sub-sample, the more accurate and precise was the elastic computation, and using the complete cylindrical sample with the new method gave still more accurate and precise results. Fortran 90 programs for the analytical solutions are made available on-line, along with the parallel finite element codes used.
The ionic diffusivity of a concrete is a function of its microstructure at many length scales, ra... more The ionic diffusivity of a concrete is a function of its microstructure at many length scales, ranging from nanometers to millimeters. The microstructure is largely controlled by the initial concrete mixture proportions and the ultimate curing conditions. Linking a property like ionic diffusivity to the microstructure then requires a multiscale approach. A multiscale microstructural computer model for ionic diffusivity has been previously developed. This model has been developed specifically to compute the chloride diffusivity of concretes with various mixture proportions and projected degrees of hydration. The three key parts of this model were dependent on large-scale supercomputer-magnitude simulations to: (1) determine the total volume of interfacial zones for a given aggregate distribution; (2) simulate the hydrated cement paste microstructure around a typical aggregate; and (3) compute the effect of the aggregates and interfacial zones on the overall diffusivity of the concrete. The key feature of this model is that one can approximately take into account the redistribution of cement paste between interfacial transition zone regions and bulk paste regions, and its important effect on overall concrete diffusivity. In the present article, we review the previously developed model and show how analytical equations can accurately replace the large scale computer simulations of parts (1) and (3). This accomplishment will make the model more usable by those who do not have access to supercomputer computing power.
The pore structure of hydrated cement in mortar and concrete is quite different from that of neat... more The pore structure of hydrated cement in mortar and concrete is quite different from that of neat cement paste. The porous transition zones formed at the aggregate-paste interfaces affect the pore size distribution. The effect of the sand content on the development of pore structure, the permeability to water, and the diffusivity of chloride ions was studied on portland cement mortars. Mortars of two water-to-cement ratios and three sand volume fractions were cast together with pastes and tested at degrees of hydration ranging from 45 to 70%. An electrically-accelerated concentration cell test was used to determine the coefficient of chloride ion diffusion while a high pressure permeability cell was employed to assess liquid permeability. The coefficient of chloride ion diffusion varied linearly with the critical pore radius as determined by mercury intrusion porosimetry while permeability was found to follow a power-law relationship vs. this critical radius. The data set provides an opportunity to directly examine the application of the Katz-Thompson relationship to cement-based materials.
The electrical conductivity of portland cement mortars was determined experimentally as a functio... more The electrical conductivity of portland cement mortars was determined experimentally as a function of the volume fraction of sand and the degree of hydration. The results were analyzed using theoretical models that represent the mortars as three-phase, interactive composites. The three phases are the matrix paste, the aggregate, and the thin interfacial transition zone between the two. The microstructure and properties of the conductive phases (the transition zone and the matrix paste) were determined by a micrometer-scale microstructural model, and were used in conjunction with random-walk algorithms and differential-effective medium theory to determine the overall mortar conductivities. The presence of the transition zone was not found to significantly affect the global electrical conductivity of the mortar. However, there were significant differences in conductivity between the transition zone and matrix pastes when examined on a local level. These differences were found to vary with hydration and were most significant when the degree of hydration was between 0.5 and 0.8.
2015 USNC-URSI Radio Science Meeting (Joint with AP-S Symposium), 2015
Electromagnetic scattering from Carbon Nanotubes (CNT) has received wide interest in the past dec... more Electromagnetic scattering from Carbon Nanotubes (CNT) has received wide interest in the past decade. Many different CNT configurations have been computationally investigated such as single CNTs, infinite planar arrays of CNTs, finite arrays with simple distributions, and bundles of CNTs. In all of the previously reported configurations, the CNTs were perfectly straight and they were arranged in a uniform distribution. However, in commercial CNT composites the CNTs typically exhibit highly complex shapes and distributions. The goal of this work is to simulate and characterize the electromagnetic scattering from multiple CNTs with realistic shapes and distributions that resemble those found in commercial composites.
ABSTRACT A numerical task of current interest is to compute the effective elastic properties of a... more ABSTRACT A numerical task of current interest is to compute the effective elastic properties of a random composite material by operating on a 3D digital image of its microstructure obtained via X-ray computed tomography (CT). The 3-D image is usually sub-sampled since an X-ray CT image is typically of order 10003 voxels or larger, which is considered to be a very large finite element problem. Two main questions for the validity of any such study are then: can the sub-sample size be made sufficiently large to capture enough of the important details of the random microstructure so that the computed moduli can be thought of as accurate, and what boundary conditions should be chosen for these sub-samples? This paper contributes to the answer of both questions by studying a simulated X-ray CT cylindrical microstructure with three phases, cut from a random model system with known elastic properties. A new hybrid numerical method is introduced, which makes use of finite element solutions coupled with exact solutions for elastic moduli of square arrays of parallel cylindrical fibers. The new method allows, in principle, all of the microstructural data to be used when the X-ray CT image is in the form of a cylinder, which is often the case. Appendix A describes a similar algorithm for spherical sub-samples, which may be of use when examining the mechanical properties of particles. Cubic sub-samples are also taken from this simulated X-ray CT structure to investigate the effect of two different kinds of boundary conditions: forced periodic and fixed displacements. It is found that using forced periodic displacements on the non-geometrically periodic cubic sub-samples always gave more accurate results than using fixed displacements, although with about the same precision. The larger the cubic sub-sample, the more accurate and precise was the elastic computation, and using the complete cylindrical sample with the new method gave still more accurate and precise results. Fortran 90 programs for the analytical solutions are made available on-line, along with the parallel finite element codes used.
The ionic diffusivity of a concrete is a function of its microstructure at many length scales, ra... more The ionic diffusivity of a concrete is a function of its microstructure at many length scales, ranging from nanometers to millimeters. The microstructure is largely controlled by the initial concrete mixture proportions and the ultimate curing conditions. Linking a property like ionic diffusivity to the microstructure then requires a multiscale approach. A multiscale microstructural computer model for ionic diffusivity has been previously developed. This model has been developed specifically to compute the chloride diffusivity of concretes with various mixture proportions and projected degrees of hydration. The three key parts of this model were dependent on large-scale supercomputer-magnitude simulations to: (1) determine the total volume of interfacial zones for a given aggregate distribution; (2) simulate the hydrated cement paste microstructure around a typical aggregate; and (3) compute the effect of the aggregates and interfacial zones on the overall diffusivity of the concrete. The key feature of this model is that one can approximately take into account the redistribution of cement paste between interfacial transition zone regions and bulk paste regions, and its important effect on overall concrete diffusivity. In the present article, we review the previously developed model and show how analytical equations can accurately replace the large scale computer simulations of parts (1) and (3). This accomplishment will make the model more usable by those who do not have access to supercomputer computing power.
The pore structure of hydrated cement in mortar and concrete is quite different from that of neat... more The pore structure of hydrated cement in mortar and concrete is quite different from that of neat cement paste. The porous transition zones formed at the aggregate-paste interfaces affect the pore size distribution. The effect of the sand content on the development of pore structure, the permeability to water, and the diffusivity of chloride ions was studied on portland cement mortars. Mortars of two water-to-cement ratios and three sand volume fractions were cast together with pastes and tested at degrees of hydration ranging from 45 to 70%. An electrically-accelerated concentration cell test was used to determine the coefficient of chloride ion diffusion while a high pressure permeability cell was employed to assess liquid permeability. The coefficient of chloride ion diffusion varied linearly with the critical pore radius as determined by mercury intrusion porosimetry while permeability was found to follow a power-law relationship vs. this critical radius. The data set provides an opportunity to directly examine the application of the Katz-Thompson relationship to cement-based materials.
The electrical conductivity of portland cement mortars was determined experimentally as a functio... more The electrical conductivity of portland cement mortars was determined experimentally as a function of the volume fraction of sand and the degree of hydration. The results were analyzed using theoretical models that represent the mortars as three-phase, interactive composites. The three phases are the matrix paste, the aggregate, and the thin interfacial transition zone between the two. The microstructure and properties of the conductive phases (the transition zone and the matrix paste) were determined by a micrometer-scale microstructural model, and were used in conjunction with random-walk algorithms and differential-effective medium theory to determine the overall mortar conductivities. The presence of the transition zone was not found to significantly affect the global electrical conductivity of the mortar. However, there were significant differences in conductivity between the transition zone and matrix pastes when examined on a local level. These differences were found to vary with hydration and were most significant when the degree of hydration was between 0.5 and 0.8.
Uploads
Papers by Ed Garboczi