Welcome to Scribd. Sign in or start your free trial to enjoy unlimited e-books, audiobooks & documents.Find out more
Standard view
Full view
of .
Look up keyword
Like this
0 of .
Results for:
No results containing your search query
P. 1


|Views: 17|Likes:
Published by Camilo Granados

More info:

Published by: Camilo Granados on Feb 22, 2013
Copyright:Attribution Non-commercial


Read on Scribd mobile: iPhone, iPad and Android.
download as PDF, TXT or read online from Scribd
See more
See less





Modeling of Chloride Transport in Cracked Concrete: A 3-D Image
 based Microstructure Simulation
Yang Lu
, Edward Garboczi
, Dale Bentz
, Jeffrey Davis
 National Institute of Standards and Technology
*Corresponding author:Engineering Laboratory, National Institute of Standards and Technology, 100 Bureau Drive8615, Gaithersburg, MD 20899, yang.lu@nist.gov 
It is highly desirable to develop amodel predicting the chloride diffusion profile incracked concrete while considering the realmicrostructure including cement paste, voids,and aggregates. While current models consider concrete at various levels of complexity in predicting the initiation of chloride-inducedcorrosion, considering the influence of crackingis generally beyond their scope. In this study, a3-D image-based microstructure simulation procedure was developed to model the chlorideingress in cracked mortar. A micro-X-rayfluorescence (XRF) test was conducted tomeasure the chloride concentration profile of amortar sample. The notched mortar sample wasexposed in a chloride ponding test for 30 days before the micro-XRF measurement. A 2-Dsimulation result, with a mesh based directly onthe XRF characterization of microstructure,showed good agreement with the micro-XRFmeasurement. With this validation, two different3-D concrete microstructures were generated andmeshed in 3-D and a commercial software package was used to accurately compute theinfluence of cracking on chloride diffusion with binding. The chloride concentration gradient inthe crack changed the concentration profile alongthe crack and nearby irregular aggregate surfacescontinually. Comparison to micro-XRFmeasurement data indicates that the contributionsof the crack play a significant role in the chlorideingress.
COMSOL Multiphysics, chloridediffusion and reaction, micro-X-rayfluorescence, X-ray CT, virtual concrete model,corrosion.
1. Introduction
To improve the service life of concretestructures and make them more durable,appropriate mechanical and transport propertiesare equally necessary. Generally, commonconcrete degradation processes, includingcorrosion of steel reinforcement, sulfate attack,alkali-silica reaction, and damage due tofreezing/thawing are strongly influenced by thediffusion of ions modified by any binding/reaction. Thus, an applicable approach toincrease concrete service life is to reducetransport phenomena in concrete. Moreover,micro- or macro- cracks, which have a criticalnegative influence on all concrete degradation processes, are almost always present in concreteduring its service life. Cracks with differentwidths and depths will reduce the effective cover thickness and accelerate the migration of chloride ions. Understanding the relationships between cracked microstructure and transportcan improve the elucidation of degradationmechanisms within concrete. While currentmodels consider concrete at various levels of complexity in predicting the initiation of chloride-induced corrosion, considering theinfluence of cracking is generally beyond their scope. This paper develops a model to predict thechloride diffusion depth in cracked concreteconsidering the real microstructure includingcement paste, voids, and aggregates.Previous research has used the COMSOLMultiphysics package
to model the microscalecorrosion and cracking process in reinforcedconcrete due to non-uniform corrosion(Pan andLu 2012). The properties of cracked and porousmedia were analyzed by COMSOL(Perko et al.).Reaction and diffusion processes in a mineralrock, without considering microstructureexplicitly, were also studied by(Richter et al.).Some studies, based on non-steady statemigration tests, were described in NT BUILD492, using an electrical field and artificial cracks(Marsavina et al. 2009).(Kwon et al. 2009) 
Certain commercial equipment and/or materials areidentified in this report in order to adequately specifythe experimental procedure. In no case does suchidentification imply recommendation or endorsement by the National Institute of Standards and Technology,nor does it imply that the equipment and/or materialsused are necessarily the best available for the purpose.
 obtained the relationship between diffusioncoefficients and crack widths in cracked concretestructures through field investigations; however,the crack depth is equally important and should be considered in the model. Past research onchloride diffusion in concrete generally focusedon specimens with artificial cracks and did notexpress the real condition and real shape of concrete cracks. Digital image processingtechnology provides an innovative method for considering the microstructure influences ontransport properties in rock and geotechnicalmaterials. Image analysis technology was used torepresent rock structures(Zhang et al.)inCOMSOL. It has also been employed to couplewith a finite element method for the two-dimensional mechanical analysis of geomaterials(Yue et al. 2003). However, there is nosystematic procedure available to simulate thecracked concrete corrosion process using a 3-Ddigital image-based microstructure model.In this study, approaches for obtaining anunderstanding of diffusion and binding (reaction)in cracked concrete using the COMSOLMultiphysics package are presented. Twodifferent 3-D microstructures with cracks wereemployed in the simulations. A 3-D X-ray CTimage-based mortar microstructure wasmeasured and a 3-D virtual microstructure usinga spherical harmonic analysis-based approach(Garboczi 2002)was created. The 3-D image- based microstructure and spherical harmonic- based microstructure both were converted into afinite element mesh form using recent techniquesdeveloped for random microstructures(Lu andGarboczi). Next, a 3-D mesh representing thetrue multiphase microstructure was written as a Nastran mesh and imported into the COMSOLMultiphysics package. Two COMSOLinterfaces, Transport of Diluted Species and User Defined ODE Mathematics, were employed inthis study to examine the influence of crackingon chloride diffusion with binding. For the purpose of validation, a micro X-rayfluoresce
nce (μXRF)
(Janssens et al. 1996)testwas performed to measure the chlorideconcentration profile of a mortar sample at NISTand compare the experimental to simulationresults. Then, the 3-D X-ray CT image-basedmicrostructure and the 3-D virtual microstructurewere both used to study the crack effect onchloride transport.
2. Image
based Microstructures
Two different representations of microstructures with cracks were employed tocompute transport and absorption binding(reaction). The microstructure study focused onconcrete composed of a matrix of cement paste(hydrated cement), sand and gravel inclusions(aggregates), and voids. The two microstructuresconsidered here are (a) X-ray computedtomography (CT) images of a mortar stackedinto a 3-D digital image, and (b) real-shapedsand and gravel particles, represented byspherical harmonic series, randomly placed intoa 3-D box to make a virtual concretemicrostructure. Based on recent 3-D meshing procedures(Lu and Garboczi), themicrostructures with a crack could be generated.
Figure 1.
Two-dimensional slices (25 mm x 25 mmmortar sample with water-to-cement ratio of 0.4) from3-D microtomographic image data set correspondingto: (a) - original X-ray CT image slice, (b) -segmented multiphase image, and (c) - crack region issegmented from other phases and the crack area isshown with a red outline. The voxel size is 0.017 mm.
The first microstructure is a 3-Dmicrostructural representation based on 2-Dimages taken at NIST, as obtained using X-raycomputed microtomography (CT). For eachsample scanned, a three-dimensionalcomputational volume is divided into a cubicgrid of voxels, where each voxel contains asingle phase (mortar, crack and voids). Figure 1 provides two-dimensional slices of the X-ray CTdata set. The specimen contains three readilydistinguishable phases based on different greylevel intensity values (or more correctly x-rayabsorption): dark air voids and empty pores, dark crack, and bright mortar (cement paste). Thesizes and shapes of many of the individual air voids are readily observed, both in their two-dimensional (Figures 1) and in their three-dimensional (Figure 2) forms. Most of theseempty pores were air voids entrained duringsample preparation. The largest connected dark 
 area is the crack phase. The crack was made by putting a stainless steel shim into the castingspecimen from the top, and then pulling it outafter final set of the mortar has occurred. Oneshould note that individual aggregates could not be distinguished from the matrix in these images.Optimizing the X-ray CT instrument in order to be able to see and segment the aggregates is anarea of future work.
Figure 2.
Three-dimensional volume rendering of crack and voids from segmented images. The 3-Dvolume rendering image indicates regions where thetop crack and voids distribution are present in thespecimen.
To perform a detailed quantitative analysis,the image set was processed and analyzed asfollows. First, to reduce the random noise present in the three-dimensional image set, eachimage was processed using a median filter. Inapplying a median filter, each voxel (imageelement) within the specimen volume is replaced by the median (grey level) value for all voxelswithin a fixed size cube centered on the voxel being considered. Median filters remove noise bysmoothing the data, while preserving smalldetails and sharp edges. After filteringoperations, the grey level histogram (a plot of thenumber of voxels containing each grey levelintensity value vs. grey level intensity) wasdetermined for this image data set. Next, theimage data set was segmented into 3 phases,voids, crack, and cement paste. Finally, the 3-Dsurface mesh and 3-D volumetric mesh werecreated from the segmented three-phase images.
(b) Multi-level 3D image (the threecolors represent matrix, crack, voids).(d) Simulated concentration contourof the X-ray CT microstructure(a) Raw 2D X-ray scanning
sequential images
(c) 3D tetra mesh imported into COMSOL,then boundary and initial conditions areapplied
3D ReconstructionCOMSOLSimulation
   M   e   s    h   I   m   p   o   r   t
Constant Cl-concentrationat the top surface
Figure 3.
Three-dimensional diffusion and bindingsimulation procedure with an X-ray microtomographyimage set; cubic specimens were 25 mm on a side.
Figure 3(a-c) shows a schematic of 3-Dreconstruction, mesh import and COMSOLsimulation for the X-ray CT image-basedmicrostructure. Figure 3(d) plots the simulationresult of the X-ray CT image-basedmicrostructure, the last step in the simulation procedure. This cracked concrete sample wasentirely covered by an (impermeable) epoxylayer except at the top surface. As shown inFigure 3, the original image data set was 2000 x2000 x 985 voxels, with each voxel representing17 µm. The intensity histogram measured oneach phase in the two-dimensional images isutilized to reconstruct a representative three-dimensional microstructure with the same voidsratio, surface area, volume, and correlation as theoriginal concrete. This original data set was usedto create a 3-D finite element mesh(Lu andGarboczi). The final 3-D tetrahedral meshcontained a total of 120689 nodes and 615166tetrahedral elements, including 573691 mortar elements, 34695 crack elements, and 6780 voidelements.

Activity (4)

You've already reviewed this. Edit your review.
1 hundred reads
Mohamed Tabet liked this
roland_korg liked this
roland_korg liked this

You're Reading a Free Preview

/*********** DO NOT ALTER ANYTHING BELOW THIS LINE ! ************/ var s_code=s.t();if(s_code)document.write(s_code)//-->