Asphalt internal structurecharacterization with X-Raycomputed tomography

Denis Jelagin, Ibrahim Onifade, Alvaro Guarin and Nicole Kringos

KTH, Highway and Railway Engineering

Outline

Understanding of asphalt mixtureproperties based on constituentmaterials spatial distribution and theirmechanical properties:

- Determination of quantitative parametersto describe asphalt internal structure.

- Mechanical modeling with finite elementmethod to quantify the impact theconstituent material parameters have onmixture mechanical behavior

Asphalt mixture internal structure and its effect on field performance

Asphalt consists of three main phases: stones, binderand air voids; their spatial distribution and propertieshave a major impact on asphalt performance:

• Stones and stone-to-stone contacts provide a primaryload carrying mechanism in compression and shear,especially at high temperatures

• Bitumen-based binder and its distribution controltensile stiffness and fracture resistance

• Air void structure controls mixture permeability,resistance to bleeding and ageing

Deficient internal structure of asphalt results in pavement failures

Pavement failures

Rutting Fatigue cracking Thermal cracking

Potholes …Blisters

X-Ray computed tomography (CT) characterization of asphalt

X-Ray CT system to acquire imageswith spatial resolution of 5-100 µm

Avizo® Fire to segment CT data and to obtainquantitative parameters for specimens structure

Use mechanical testing to investigatethe impact of the observed internalstructure on materials performance

FEM modeling to quantify theeffect of different micromechanicaland geometrical parameters onmaterials performance

PreprocessforFEA

X-Ray CT characterization of asphalt

• Porous (“quiet”) asphalt- Cylindrical core 80 mm high x 100 mm

diameter- High air voids (20%) to facilitate drainage

and noise damping

• CT data with 59x59x59 µm voxel size is acquired

• Analysis is performed on a rectangularvolume (60x60x40 mm) in the center ofthe specimen

Analysis procedure

X-Ray CT slice beforepost-processing:

• Significant densityvariation within phases(stones and binder)

• Considerable amount ofbeam hardening

• Image noise

Analysis procedure

Corrected image:

• Histogram equalization to improve contrast

• Noise reduced with medianfilter (3x3 kernel) and edgepreserving smoothing filter

• Beam hardening corrected based on background flat field correction- Illumination profile:

Analysis procedure

Segmented image:• Phase (air voids and

stones) identification withthreshold-basedsegmentation

• Binder is defined as thedifference between totalvolume, stones and airvoids

• Stones are separated basedon the distance map withwatershed segmentation

• Stones smaller than 2.34mm are filtered out andreplaced with binder

ResultsStone skeleton

Reconstructed stone surfacesParameters describing stone sizedistribution, their shape , roughness andorientation in the material are obtained.These parameters define to a greatextent the stone skeleton strength andits susceptibility to aggregate breakage.

ResultsStone skeleton

ResultsStone skeleton (contact regions)

During separation based onthe distance map, the contactregions between stones areidentified:• Regions where the separation

lines intersect with thesegmented stone phaserepresent contact regions

• A sensitivity range forcontact detection is definedpresently as 108 µm (2pixels)

ResultsStone skeleton (contact regions)

The stone contact regions provide aprimary load transferring mechanism incompression and shear.In several recent studies contact zonesgeometry and orientation have beencorrelated with asphalt compactabilityand rutting performance.

ResultsAir voids

CT data is analyzed in order to evaluate ifthe air void distribution and connectivityin the specimen agree with the designparameters of the mixture.Reduced air void content at the bottom ofthe specimen results in compromisedpermeability and noise dampingcapabilities.

Reconstructed air voids surfaces

Micromechanical analysis with FEM

FEM simulations based on structural information obtained with the X-Ray CT allow to:

• Improve our understanding of the mechanical behavior of asphalt and its degradation processes.

• Quantify the effect of using constituent materials with improved (or worsened) characteristics.

• Develop a “virtual specimen” type of approach for asphalt mixturedesign. This will provide a cost effective way to optimize differentasphalt mixture parameters, e.g. binder type, air void contents andstone size distribution for better field performance.

Analysis results illustrate the capability of this method tocapture stress concentrations and strain localization arisingdue to differences in mechanical and thermal properties of thephases.

Uniaxial tension and thermal stresses(2D)

• Reconstructed surfaces andvolumes are exported toCOMSOL Multiphysics package

• Mechanical and thermalproperties representative foreach phase are assigned tostone and binder regions in themodel

• 2D plane strain analysis for:- Uniaxial tension- Thermally induced stresses

(temperature at the air voidboundary is reduced at a rate of10ºC/hour)

h=0.1 mm

Uniaxial tension (2D)

• Strains are localized in thebinder phase

• Strains up to 12% are observedas compared to approx. 0.2%predicted for homogeneousmaterial case

• The information obtained withthis type of modeling can beused to identify representativestress and strain levels forbinder testing

Uniaxial tension (2D)

• Load transfering chainscan be seen in thematerial

• Only main loadtransfering regions inthe binder are subjectedto a tensile stress>10MPa (as comparedto the uniform tension of19 MPa for the uniformmaterial case)

Thermal stresses (2D)

• Temperature variation ofapprox. 1.5ºK can be seen.The temperature gradientwould increase withincreasing cooling speed anddecreasing air void content.

• As the specimen is notconstrained, this type ofthermal loading would resultonly in negligible stresses inthe homogeneous material.

Thermal stresses (2D)

• Stones are subjected tohigher stresses due to theirhigher stiffness

• Regions of localized tensionare formed in the binder dueto difference in thermalcontraction propertiesbetween phases.

• Maximum tensile stresses in the binder reach approx. 2.5 MPa

Uniaxial compression (3D)

Analysis of small regionsaround stone-to-stonecontact zones to get insightinto the local degradationmechanisms:

- Work in progress…

Uniaxial compression (3D)

Von Mises stress localized in stones aroundcontact points

Compressive strains localized in the binder

Understanding the mechanisms controlling:• stone breakage and polishing during asphalt

compaction• Micro-fracture initiation in binder films

Thank you for your attention