Evaluation of Asphalt Field Cores with Simple Tomography

Evaluation of Asphalt Field Cores with Simple Tomography

Evaluation of Asphalt Field Cores with Simple

Performance Tester and X-ray Computed

Tomography

Licentiate Thesis

Florentina Angela Farca ș

Division of Highway and Railway Engineering

Department of Transport Science

School of Architecture and the Built Environment

Royal Institute of Technology

SE-100 44 Stockholm

TRITA-TSC-LIC 12-002

ISBN 978-91-85539-82-6

Stockholm 2012

ABSTRACT

The importance of aggregate structure and air voids distribution for asphalt mixture rutting and cracking performance has been well established on the basis of experience and is well documented in the literature. Past and current investigations are limited to assessment of performance based on macroscopic behavior due to the difficulty associated with the quantitative measurement and analysis of the internal structure of asphalt mixtures. Lately, technical advances in X-ray Computed Tomography (CT) and image processing and analysis has made possible to bring the attention also to the internal structure of asphalt mixtures.

SPT results from asphalt field cores, including dynamic modulus (before and after loading) and microstrain accumulation (flow number), exhibited significant variability; most likely, induced by irregularities in the core shape. The analysis of aggregate structure and air voids distribution performed trough X-ray CT, clearly identified segregation in the asphalt mixture as a key factor that induced variability in SPT results.

X-ray CT provides fundamental resources to enhance understanding about role that aggregate structure and air voids distribution of asphalt mixtures play on rutting and cracking of asphalt mixtures; such valuable knowledge could eventually generate further development of asphalt mixture design procedures and/or optimization of pavement construction methods that ultimately may lead to long lasting and economical asphalt pavements structures.

Keywords: asphalt mixture, dynamic modulus, flow number, simple performance tester, X-ray computed tomography, air voids distribution.

Acknowledgements

First of all, I would like to express special thanks to my supervisor professor

Björn Birgisson for providing great assistance and encouragement during my thesis.

I would also like to thank the following people:

• Co-supervisors: Dr. Alvaro Guarin and Dr. Denis Jelagin

• Guest professor: Manfred Partl

• Associate professor: Nicole Kringos

• Skanska: Kenneth Olsson

• NCC: Dr. Jonas Ekblad

• Colleagues: Agneta Arnius, Åsa Laurell Lyne, & others

• Family and Friends

• All others that I might have forgot to mention

Florentina Farcaș

Stockholm, March 27, 2012

i

Table of Contents

Chapter 1 Introduction

........................................................... 1

1.1

1.2

1.3

Background ............................................................................................ 1

Objectives ............................................................................................... 2

Scope ....................................................................................................... 2

Chapter 2 Literature Review

.................................................... 3

2.1

2.2

Simple Performance Tester .................................................................. 3

X-ray Computed Tomography ............................................................. 5

2.2.1

Microstructure Characterization .............................................................. 7

2.2.2

Air Void Distribution............................................................................. 10

2.2.3

Permeability analysis ............................................................................. 11

2.2.3.1

Microstructure Evolution and Damage during Loading .......... 12

Chapter 3 Laboratory Testing

.................................................. 15

3.1

3.2

Field Cores ........................................................................................... 15

Testing Equipment .............................................................................. 16

3.2.1

Simple Performance Tester (SPT) ......................................................... 16

3.2.1.1

Dynamic Modulus .................................................................... 19

3.2.1.2

Flow Number ........................................................................... 21

3.2.2

X-Ray Computed Tomography (CT) System ........................................ 24

3.2.2.1

Warm up the system................................................................. 27

3.2.2.2

Scan the specimen .................................................................... 27

3.2.2.3

Calibrate the system ................................................................. 28

3.2.2.4

Reconstruction ......................................................................... 30

3.2.2.5

Post-processing ........................................................................ 31

Chapter 4 Test Results

.......................................................... 33

4.1

4.2

Field Core Samples .............................................................................. 33

Simple Performance Tester Results ................................................... 34

4.2.1

Dynamic Modulus Test ......................................................................... 34

4.2.2

Flow Number Tests ............................................................................... 41

4.3

X-ray Computed Tomography Results .............................................. 45

4.3.1

Air Void Distribution............................................................................. 46

4.3.1

Aggregate Structure ............................................................................... 52

Chapter 5 Closure

................................................................ 55

5.1

5.2

Conclusions .......................................................................................... 55

Recommendations ................................................................................ 56

Bibliography

......................................................................... 57

Appendix

......................................................................... 73

A

B

Simple Performance Tester (SPT) ......................................................... 73

X-ray CT Results ................................................................................... 85

iii

Table of Figures

Figure 1. Gradation chart. ........................................................................................ 16

Figure 2. The IPC SPT machine. ............................................................................. 17

Figure 3. IPC SPT computer software running dynamic modulus test. ................... 17

Figure 4. Gauge Point Fixing Jig. ............................................................................ 18

Figure 5. SPT - sample setup. .................................................................................. 19

Figure 6. Dynamic Modulus - schematic loading. ................................................... 20

Figure 7. Flow Number Test - schematic loading. ................................................... 21

Figure 8. Cumulative Permanent Strain vs. Load Cycles. ....................................... 22

Figure 9. Repeated Load Test principle - schematic of flow number test loading... 24

Figure 10. The CT machine X-5000. ....................................................................... 25

Figure 11. Attenuation coefficients for different materials [90]. ............................. 26

Figure 12. General mechanism of X-ray tomography scanning. ............................. 26

Figure 13. X-ray CT - sample setup......................................................................... 28

Figure 14. The calibration rod. ................................................................................ 29

Figure 15. Images of the calibration tool. ................................................................ 30

Figure 16. Dynamic Modulus - sample rotation. ..................................................... 34

Figure 17. Dynamic Modulus - different loads and misalignment. ......................... 35

Figure 18. Dynamic Modulus [ksi] vs. Frequency [Hz]. ......................................... 36

Figure 19. Dynamic Modulus [ksi] vs. Frequency [Hz] - logarithmic scale. ........... 36

Figure 20. Dynamic Modulus [ksi] vs. Reduced Frequency [Hz]. .......................... 37

Figure 21. Phase Angle [deg] vs. Reduced Frequency [Hz]. ................................... 37

Figure 22. Master Curve comparison (Test results, Oscarsson and Nilsson). ......... 38

Figure 23. Phase Angle comparison- (Test results, Oscarsson and Nilsson). .......... 39

Figure 24. Comparison of Dynamic Modulus for different labs at a reference temperature of 20°C. ....................................................................................... 40

Figure 25. Data comparison - stem-and-leaf. ........................................................... 40

Figure 26. Flow Number for different samples........................................................ 42

Figure 27. Flow Number Test Result for Sample FN2. ........................................... 42

Figure 28 Creep curve - test stages. ......................................................................... 43

Figure 29. Microstrain Samples CT1 to CT4. .......................................................... 44

Figure 30. Air voids distribution for sample CT1 before and after loading............. 47

Figure 31. Air voids distribution for sample CT2 before and after loading............. 47

Figure 32. Air voids distribution for sample CT3 before and after loading............. 48

Figure 33. Air voids distribution for sample CT4 before and after loading............. 48

Figure 34. Air voids size histogram before and after loading for samples CT1 and

CT4. ................................................................................................................. 50

Figure 35. Air voids size histogram before and after loading for samples CT2 and

CT3. ................................................................................................................. 51

Figure 36. X-ray CT image. ..................................................................................... 52

Figure 37. Distribution of aggregates in a 3D volume of interest of sample CT1. .. 53

Figure 38. Example of aggregate size classification in the 3D volume of interest. . 54

iv

Figure 39. Dynamic Modulus [ksi] vs. Reduced Frequency [deg] shifted - at different temperatures. .................................................................................... 81

Figure 40. X-ray CT image - sample CT3. .............................................................. 85

Figure 41. Aggregates in the 3D volume of interest- sample CT1. ......................... 87

v

Index of tables

Table 1. Characteristics for Dynamic Modulus tests. .............................................. 21

Table 2. The samples. .............................................................................................. 33

Table 3. Flow Number - test results. ........................................................................ 41

Table 4. Samples CT1-CT4 – Dynamic Modulus and Load Cycles. ....................... 44

Table 5. Volume resolution. .................................................................................... 45

Table 6. Total air voids content for samples CT1-CT4 before and after loading. ... 49

Table 7. Flow Number Assumptions and SPT system features. .............................. 77

Table 8. Data quality statistics requirements for Dynamic Modulus measured with

SPT. ................................................................................................................. 77

Table 9. Statistical analysis of influence in sample rotation. ................................... 77

Table 10. The SPT results and Master Curve calculation. ....................................... 83

Chapter 1

Introduction

1.1 Background

Aggregates are a key constituent of asphalt mixtures, since they represent about

95% of the total weight of the mixture. Particle size distribution and air voids distribution are factors that affect the most important properties of asphalt mixtures, such as rutting and cracking resistance.

The SPT for Superpave mix design was developed as part of the National

Cooperative Highway Research Program NCHRP project 9-29 “Simple

Performance Tester for Superpave Mix Design”, from 2001 until 2011. Under this project the Advanced Asphalt Technologies research agency proof-tested

SPT for permanent deformation and fatigue cracking in asphalt mix design. SPT has been mostly used for laboratory-prepared asphalt samples; this project evaluated the potential use of field cores for SPT

X-Ray Computerized Tomography (X-Ray CT) is a non-destructive technique that allows visualizing the interior of solid objects by capturing digital information on their 3-D microstructure. Several researchers have used this equipment for construction materials to characterize internal structure, determine air void distribution, quantify material permeability, and determine the evolution of microstructure during loading.

This work was focused on evaluating the viability of using SPT for characterization of asphalt field cores and the use of X-ray CT for evaluation of the internal structure of asphalt mixtures e.g., aggregate structure and air voids distribution.

2 Chapter 1 Introduction

1.2 Objectives

The main purpose of this study was to investigate variability of SPT results when asphalt field samples are evaluated; X-ray CT was also utilized to analyze the internal structure (Aggregate structure and air voids distribution) of the asphalt cores and possibly identify factors that may have induced variability of

SPT results. Detailed objectives of this research work were as follows:

Simple Performance Tester SPT

• Measure dynamic modulus

• Generate dynamic modulus master curve

• Evaluate variability of dynamic modulus

• Determine flow number

• Establish characteristic creep curve

X-ray CT

• Evaluate air void distribution along the vertical direction

• Assess air voids distribution before and after loading

• Compare aggregate structure from different samples

• Estimate particle size distribution

1.3 Scope

This study focused on the use of asphalt field cores to perform Simple

Performance Tester SPT and in the detection of elements that can induce variability on SPT results. An X-ray CT system was employed for detailed analysis of the internal structure of the asphalt mixture. All the field cores were produced using a single asphalt base type (AG22) prepared with binder type

70/100.

Chapter 2

Literature Review

2.1 Simple Performance Tester

Simple Performance Tester (SPT), also widely known as the Asphalt Mix performance Tester, is a computer controlled hydraulic testing machine to measure dynamic modulus and flow number of asphalt mixtures according to the AASHTO TP79 specification.

A project sponsored by the Federal Highway Administration began in 1996 at the University of Maryland to validate SPT for measurement of permanent deformation and fatigue cracking. Three years later, National Cooperative

Highway Research Program (NCHRP) Project 9-19 [1] targeted the finalization

of protocols for SPT to be incorporated in the Superpave volumetric mix design method. A final report was prepared including updated test protocols for the validated SPT and guidelines for their implementation and adoption by

AASHTO.

The SPT for Superpave mix design was part of the project NCHRP 9-29

“Simple Performance Tester for Superpave Mix Design”, [2] from 2001 until

2011. Under this project the Advanced Asphalt Technologies research agency was assigned the task of designing, procuring, and evaluating an SPT for:

• Proof-testing for permanent deformation and fatigue cracking in HMA mix design and

• Materials characterization for pavement structural design according to the Mechanistic-Empirical Pavement Design Guide (MEPDG).

The phase I of NCHRP project 9-29 in 2002 [3], evaluated several tests for

permanent deformation, fatigue cracking, and low-temperature cracking. This

4 Chapter 2 Literature Review

report recommended three test-parameter combinations for further field validation:

• Dynamic modulus , E*/sinφ,

• Flow time, Ft, determined from the triaxial static creep test; and

• The flow number, Fn, determined from the triaxial repeated load test.

The Phase II of the project generated NCHRP Report 513 [4] in 2003, that

analyzed the variability of dynamic modulus and flow number. The Phase III,

NHCRP Report 530, led to the production of a SPT machine capable of accurately measuring dynamic modulus for calculation of master curves.

In 2004, the NCHRP report 530, “Evaluation of Indirect Tensile Test (IDT)

Procedures for Low-Temperature Performance of Hot Mix Asphalt” [5];

underlines the importance of understanding of low-temperature cracking mechanisms in asphalt pavements and contributes to SPT development by reducing test variability and improving its precision and reliability.

The Phase IV of the NCHRP project report 614, [6] “Re

fining the Simple

Performance Tester for Use in Routine Practice” proposed a new standard practice for developing the dynamic modulus master curve (frequency and temperatures for testing) for a limited temperature range (from 4°C to 40°C).

Improvements for cooling capacity, load capacity, indicators were also suggested.

NCHRP Phase V report 629 [7] “Ruggedness Testing of dynamic modulus and

flow number tests with the Simple Performance Tester”, describes a series of experiments to be conducted and analyzed to assess the SPT equipment and test procedures for the dynamic modulus and flow number tests. Phase V included two major experiments:

• A formal ruggedness experiment in accordance with ASTM E1169,

Standard Guide for Conducting Ruggedness Tests.

• An experiment designed to investigate whether there are significant differences in SPT data collected with equipment from the three manufacturers: Interlaken Technology Corporation (ITC); IPC Global

(IPC); and Medical Device Testing Services (MDTS).

Phase VI (2011) NCHRP 702 report was available as, [8] “Precision of the

dynamic Modulus and flow number tests conducted with the asphalt mixture performance tester”. An inter-laboratory study was designed to analyze:

• Dynamic modulus and phase angle,

• Unconfined flow and

• Permanent strain in confined flow number tests.

X-ray Computed Tomography 5

The findings of this project were multiple:

• Variability in the tests increases with decreasing specimen stiffness. The variability of low stiffness dynamic modulus tests and the permanent deformation in confined flow number tests is higher.

• Variability of unconfined flow number tests is unacceptable considering

current criteria for rutting resistance (NCHRP Project 9-33) [9].

• Specimen fabrication was found to be a major source of between-lab variability in both the dynamic modulus and flow number tests.

Compactor type, air void content, and specimen age were evaluated, and none were found to have a systematic effect on the study dynamic modulus and flow number data.

• Gauge point drift was evident in the high-temperature dynamic modulus test data from two of the participating laboratories out of 7.

• Differences in the fabrication and use of the greased latex end friction reducers are likely a source of significant variability in the flow number tests. Better control on the type, amount, and distribution of the grease is needed.

The NCHRP Project 9-29 successfully devised tests, methods and specifications for the development of a SPT machine. The project resulted in the development, improvement and validation of SPT machines by several manufacturers.

2.2 X-ray Computed Tomography

X-Ray Computerized Tomography (X-Ray CT) is a non-destructive technique that allows visualizing the interior of solid objects by capturing digital

information on their 3-D microstructure [10]. X-Ray CT consists generally of

an X-Ray source, a detector, and a turntable carrying the test specimen in between the source and the detector. X-Rays intensities are measured before and after they are emitted through the specimen in different directions for a full rotation of the specimen. The intensity values are used to calculate the distribution of the linear attenuation coefficient in order to generate a map representing the density at every point in the microstructure. Brighter regions correspond to dense objects such as aggregates, and dark regions correspond to low-density objects such as voids. X-Ray CT systems are very sensitive to small variation in density, which could be as low as 1% or smaller. This enables the

X-Ray CT system to characterize a wide spectrum of engineering materials (e.g.

[11], [12], [13], [14]).

For a long time, researchers have assumed that granular construction materials such as asphalt and concrete are isotropic in developing continuum models. In reality, these materials are complex composite structures of aggregates (rock)

6 Chapter 2 Literature Review

and binder material consisting of bitumen and cement paste for asphalt and concrete, respectively. There is a vital need to quantify the relationship between microstructure characteristics that are known to influence the behavior of the

material and macroscopic properties of interest to engineers and designers [15].

X-Ray CT systems provide the ability to characterize the microstructure and its evolution for asphalt and concrete as these materials are subjected to loading, which is critical for a better understanding of the behavior of these composite construction materials needed for the development of more realistic prediction models of the performance of these materials. In fact, without a clear understanding of the evolution of the microstructure, the understanding of the

deformation process is limited [16]. However, most of the available continuum

models for asphalt and concrete materials are developed without experimental measurements of the microstructure distribution. This is due to the difficulties associated with the quantitative analysis of the microstructure, which has prevented continuum modeling from becoming a state-of-the-practice technique

for composite construction material engineering applications [17].

Recently, there have been several successful attempts to quantify the microstructure of granular materials using imaging technology. Initial attempts focused on two-dimensional (2-D) measurements conducted on cut sections of

the material (e.g. [12], [13], [18], [19], [20]). The 2-D measurements were extended to 3-D using stereological principles [21]. The stereology approach to quantify the microstructure was initiated by Hilliard in the 1960’s [22], [23] and expanded to Cartesian tensor formulation by Kanatani in 1980’s [24], [25], who

presented a systematic approach using Buffon transform and microstructure tensors to represent the distribution of microstructure quantities. However, this approach is laborious, destructive, ineffective in capturing the evolution of the microstructure, and provides an approximation of the actual 3-D distribution from 2-D measurements.

X-Ray computed tomography (CT) is fast becoming a powerful tool to accurately and non-destructively characterize the microstructure of many granular materials. It has been successfully utilized to quantify the 3-D

microstructure of asphalt (e.g. [14], [26], [27], [28], [29], [30], [31], [32], [33],

[34]), granular soils (e.g. [35], [36]), and cement concrete (e.g. [15], [37]). X-

Ray CT images can be processed using image processing techniques to digitally reconstruct the 3-D microstructure of the scanned specimen. A key aspect of X-

Ray CT equipment is that it allows for further mechanical testing of the specimen after initial loading and imaging, where one can relate the microstructure to the mechanical response of the material.

X-ray Computed Tomography 7

X-Ray CT equipment has been used for construction materials to:

• Characterize their microstructure.

• Determine air void distribution.

• Quantify material permeability.

• Determine the evolution of microstructure during loading.

2.2.1 Microstructure Characterization

Synolakis et al. (1996 [26]) presented a new method for computing the

microscopic internal displacement fields associated with the permanent deformation of 3-D asphalt cores while satisfying the small gradient approximation of continuum mechanics. They computed the displacement field associated with diametral loading of a cylindrical asphalt core using X-Ray CT to collect 3-D images from sequences of 2-D images scanned before and after loading. The pair of 3-D images was then used to compute the displacement field by comparing their 3-D representation before and after the deformation.

Yue et al. (1995 [38]) showed the ability of image analysis techniques to

capture some aspects of the internal structure of asphalt. Masad et al. (1999

[27]) used this technology to measure the aggregate orientation and aggregate

gradation of asphalt. Images captured using X-Ray CT were used to analyze air void distribution and images captured were used to study aggregate orientation and segregation. Segregation refers to preferential separation of coarse and finegraded aggregates within the material, leading to reduced life and durability problems. Segregation can be caused by the material design, improper handling, or compaction. It is therefore important to be able to quickly determine the presence of segregation through X-Ray CT scanning.

Hunter et al. (2004 [39]) studied the internal structure of asphalt formed by

different compaction methods including the gyratory compactor, vibratory, and slab compaction by slicing the material into thin sections. They found that the circumferential aggregate orientation increased with increasing particle size in the gyratory compactor and vibratory compactors. Aggregate segregation was also found to differ using different compaction methods. They also conducted a repeated load axial test and found that the gyratory compactor and vibratory compacted specimens showed higher resistance to permanent deformation than slab compacted specimens.

Tashman et al. (2001 [40], [41]) and Birgisson et al. (2005 [42]) developed

automated image processing algorithm to isolate the aggregates from the other phases in a digital image and separate those that are in contact. The algorithm significantly improved the accuracy of the image analysis to determine the orientation, segregation, and gradation of aggregates. In their study, Tashman et

8 Chapter 2 Literature Review

al. (2001 [13]) conducted a comparison between the microstructure

measurement of laboratory compacted specimens and field cores from asphalt pavements. The study showed significant difference in terms of aggregate orientation, segregation, and air void distribution, i.e., laboratory prepared asphalt specimens did not simulate the field condition in terms of the microstructure distribution. Segregation was noticed in laboratory compacted specimens. Aggregates had a more preferred orientation towards the horizontal direction (perpendicular to the applied load) in field cores than in laboratory compacted specimens. This has a major impact on interpreting experimental results from laboratory prepared asphalt specimens to predict field performance.

In addition, it was found that laboratory compacted specimens exhibited axisymmetric aggregate distribution, where the aggregates had a preferred orientation in sections cut vertically but had random distribution when the sections were cut horizontally.

The effects of anisotropy on modulus and strength of construction materials need to be studied. Only few studies have been conducted to establish a relationship between these key engineering properties and the material

microstructure. Masad et al. (2002 [43]) used image analysis techniques to

study the modulus anisotropy of asphalt mixtures within the framework of a micromechanics-based model. They found that the stiffness in the vertical

(axial) direction is 30% more than that in the horizontal (lateral) direction. This

was consistent with the finding of Tashman et al. (2001 [13]) on the

axisymmetric distribution of asphalt microstructure.

Similarly, microstructure anisotropy causes a coupling between the volumetric and deviatoric response. This coupling effect is an important feature in modeling the behavior of granular materials, for which inelastic dilation is a

dominant effect [44]. Granular materials generally exhibit anisotropic

microstructure distribution so that a scalar quantity is not sufficient to

characterize it (i.e. [13], [18], [32], [45], [46], [47], [48], [49], [50]). Therefore,

it is necessary to introduce microstructure quantities that can represent the directional nature of the microstructure. These quantities are referred to as

“microstructure tensors” and are determined from measurements on the solid or void phase.

Tomographic imaging systems have been used to evaluate existing microstructure tensor formulations and possibly develop new formulations based on the scanned images obtained from a wide range of asphalt and concrete materials. The most popular microstructure quantities are the aggregate contact normals, aggregate orientation, void/crack orientation, and branch

vectors (e.g. [25], [46], [48]). The contact normal is defined as the vector

normal to the tangent plane at the point of contact between aggregates. An aggregate orientation is defined by the direction of its longest axis. A branch

X-ray Computed Tomography 9 vector is represented by a line joining the centers of mass of the contacting aggregates. The directional distribution of voids is described by dividing the void space into a number of “unit voids”, and assigning a vector to describe the

orientation of each unit void [51].

A continuum representation of the anisotropic distribution of the microstructure is achieved by averaging the directional distributions of the different microstructural quantities within a representative volume element. Kanatani

(1984 [24], 1985 [25]) presented a stereological-based approach using Buffon

transform and microstructure tensors to describe the directional distribution of microstructure quantities, regardless of the quantity under consideration.

Kanatani (1985 [25]) and Muhunthan and Chameau (1997 [49]) showed that the

anisotropy of engineering materials can be approximated using only a second order tensor.

Past efforts relied on destructive techniques that involved cutting specimens in equally spaced sections parallel to three orthogonal planes and applying stereological principles along with the assumption of randomness to obtain the

3-D distribution from the 2-D measurements (e.g. [19], [24], [25], [52]). Using

such an invasive technique restricted the applicability of the above procedure in capturing the evolution of the microstructure during deformation. Its main shortcomings are the large number of samples required for taking accurate successive measurements and the bias introduced due to the randomness assumption. The X-Ray CT offers a solution to such a problem through 3-D measurements of the distribution function in order to determine the components of the deviatoric microstructure tensor.

Characterizing the microstructure of asphalt and concrete requires the isolation of the individual phases in the first place before conducting any image analysis.

These phases are the aggregates, air voids/cracks, and the binder for asphalt or for concrete. The aggregate contact normal distribution is another important

microstructure quantity. Watson et al. (2004 [53]) used a new technique to

verify the voids in coarse aggregate (VCA) concept for defining stone-on-stone contact in open graded friction coarse asphalt mixtures. The image analysis technique had the advantage over the VCA method in that it can determine the number of contact, which is related to stiffness, while the VCA method gives

only a “yes” or “no” answer to whether the stone-on-stone contact exists [53].

A statistical parameter (

) that can be used to quantify the directional distribution of aggregate orientation or contact normals was developed by

Curray (1956 [54]); theoretically, the value of

∆ ranges between zero and unity.

Zero value indicates the aggregates are completely randomly distributed, which is analogous to isotropic materials, and a unity value indicates the aggregates

(or the contact normals) are all oriented in one direction.

10 Chapter 2 Literature Review

Oda and Nakayama (1989 [47]) derived a microstructure tensor that describes

the aggregate orientation (or contact normal) in terms of a vector magnitude that indicates that aggregate distribution in HMA is anisotropic and that the aggregates have a preferred orientation toward the direction perpendicular to the direction of the applied load. This was illustrated for two asphalt mixtures by

Tashman et al. (2001 [29]), who showed that

ranges between 0.3 and 0.5 on vertical sections of HMA, whereas it does not exceed a value of 0.1 on horizontal (lateral) sections. X-Ray CT systems have been used to further evaluate aggregate particle distribution for a wider range of asphalt and concrete materials.

2.2.2 Air Void Distribution

Though air voids in concrete and asphalt possess no appreciable mechanical strength, their distribution is important in determining the overall response of

the material [16]. Wang et al. (2001 [17]) stated that a statistical study of the air

void size and spatial distribution in asphalt would present valuable information leading to a better understanding of the permanent deformation and fatigue

mechanisms of asphalt. In their work, Wang et al. (2001 [17]) studied the spatial

and size distribution of the air voids in different asphalt field specimens with known permanent deformation performance. They used X-Ray CT imaging and a virtual cutting technique to conveniently obtain the cross-sections in different orientations.

Masad et al. (1999 [12]) found that air void distribution in gyratory compacted

laboratory asphalt specimens exhibit a “bath-tub” shape where more air voids were present at the top and bottom parts of a specimen. This shape was more pronounced at higher compaction efforts. They also found that specimens prepared with different aggregate sizes were found to have noticeably different air void sizes.

Masad et al. (2002 [31]) and Birgisson et al (2005 [42]) used X-Ray CT along

with image analysis techniques to characterize the statistical distribution of air void sizes at different depths in asphalt specimens; they found that air voids follow a Weibull distribution. About 40% of the total number of air voids was found to concentrate at the top third of the sample. In the case of specimens prepared using linear kneading compactor, air void content was found to increase with depth. The effect of gradation was also well reflected on the air void size; the coarser gradation showed larger air voids. In summary, the results mean that different compactors used in current practice produce compact asphalt that can have a significantly different microstructure and thus also a different loading response. There is currently a strong need to better understand

X-ray Computed Tomography 11 the relationship between internal microstructure of asphalt and its: loading response, durability under loading, extended environmental exposure.

2.2.3 Permeability analysis

One of the most promising implications of characterizing the 3-D air void distribution in HMA is the ability to identify and distinguish the connected air voids from the total air voids. This is very important for accurate characterization of asphalt permeability as it is related to the connected air voids, whereas the isolated air voids do not contribute to this phenomenon.

Al-Omari et al. (2002 [55]) developed an imaging algorithm that compares the

location of air voids in successive X-Ray images taken along the height of asphalt specimens. The algorithm identifies if a void in an X-ray CT image overlaps with another in the image underneath it. The algorithm retains the overlapping voids and deletes the ones that are not, thus isolating the connected air voids along the entire depth of the specimen. After the connected air voids had been isolated, several information were obtained that were related to the permeability characteristics including the total effective void content, specific surface area of the voids, and tortuosity (from the center of masses of the connected voids). These parameters were related using the Kozeny-Carman

equation to determine the permeability as in Walsh and Brace (1984 [56]).

Masad, Birgisson, Al-Omari, and Cooley (2004 [57]) used X-ray CT imaging to

establish air void and permeability gradients in ten field sections. Subsequently, they used these gradients in an unsaturated flow finite element model to evaluate the impact on the ingress and flow of water through these pavement sections.

Tashman et al. (2003 [58]) developed a finite difference numerical simulation

for fluid flow in granular materials by solving the continuity equation and momentum equations (x- and y- direction) for every pixel within the microstructure using a non-staggered scheme arrangement. The non-staggered scheme allows using the same finite difference grid for the continuity cells, momentum cells in x-direction, and momentum cells in y-direction. Hence, each pixel in the digital image of the microstructure represents the continuity cell as well as the momentum cell in both directions.

Birgisson, et al. (2005 [42]) similarly developed a user-based subroutine within

a commercial 3-D finite element code to simulate the flow of water through scanned images of asphalt specimens. This model was used by Birgisson, et al.

(2005 [42]) to evaluate a new moisture conditioning procedure developed for

the Florida Department of Transportation and the Federal Highway

Administration. Birgisson et al. (2005 [42]) and Castelblanco, Masad, and

12 Chapter 2 Literature Review

Birgisson (2005 [59]) used X-Ray CT 3-D images of asphalt specimens with

known moisture damage potential, along with calculated permeabilities determined from X-Ray CT imaging to illustrate the effect of aggregate gradation on moisture damage.

2.2.3.1 Microstructure Evolution and Damage during Loading

Landis and Keane (1999 [15]) used a high-resolution X-Ray microtomography

to measure internal damage and crack growth in small mortar cylinders loaded in uniaxial compression. In their experiment, small mortar cylinders were inserted into a small loading frame that could be mounted directly on the X-Ray rotation table. This was done in order to scan the specimens at varying strain values so that the internal damage could be quantified and correlated with load deformation information. Multiple tomographic scans were made of the same specimen at different levels of deformation applied through a custom built loading frame, and image analysis of the scanned images was used to measure the internal crack growth during each deformation increment. They showed that under monotonic loading of concrete, there was elastic deformation up to 30% of peak load; beyond this point, cracking occurred at the cement-aggregate interface. At about 70% of the peak load, these distributed cracks started to localize and matrix cracking occurred, which macroscopically became largescale axial splitting. Post-peak response was characterized by additional matrix cracking and frictional mechanisms in a relatively narrow band.

Tashman et al. (2004 [34]) conducted a study where asphalt specimens were

scanned initially then deformed to prescribed strain levels of 1%, 2%, 4%, and

8% in a triaxial test set-up. The test was stopped when the prescribed strain was achieved, and the deformed specimens were imaged again. They found that the asphalt specimens illustrated a clear localization behavior that appeared related to the microstructure of the material used.

Using a particle representation approach Wang et al., (2003 [60]) studied the

evolution of the aggregate structure of asphalt subjected to a permanent

deformation test. Wang et al. (2005 [61]) quantified the particle displacement

through obtaining the differences of the particle mass center coordinates before and after testing, which allowed the determination of the average strain in a small element consisting of four adjacent particles. The strains in the surrounding mastic were quantified assuming the aggregate particles have only rigid motions. The study indicated that the strains at the microstructure level deviate significantly from the strains computed through homogeneous continuum theories, and that the strains in the mastic could be ten times larger than the average strains. Nevertheless, the overall average of these strains resulted in the same displacements observed at the boundaries. The experimental observations from the limited study performed by Wang et al.

X-ray Computed Tomography 13

(2005 [61]) have two significant implications: a) the properties of the binder and

mastic at larger strains need to be characterized for a better description of the mixture properties, and b) the binder and mastic properties at small strains may not represent their behavior at larger strains.

In order to study plastic deformation it is necessary to define a yield surface for granular materials; many plasticity models dealing with anisotropic materials

have been developed, especially for soils (e.g. [20], [46], [47], [48], [62], [63],

[64], [65]). One of the methods of incorporating the microstructure in a

continuum model for an anisotropic material is by replacing the stress tensor with a combined tensor that consists of a stress tensor and a microstructure

tensor [45], [46], [66].

The material deformation represented by the shear strain rate tensor, which is the deviatoric part of the strain rate tensor, can be related to the rate of change of microstructure tensor considering that both are deviatoric through the use of a

tensor valued functional representation as in Boehler (1987 [67]). This approach

allows for the quantification of the changes in asphalt microstructure as reflected by the deviatoric microstructure tensor D

ij

, and relates it to the macroscopic strain of the material as it undergoes the deformation process.

The literature has shown that X-Ray CT is a powerful tool to characterize and capture the damage within a material microstructure. Its power stems from the fact that it is non-destructive; hence the tested specimen is still intact for further mechanical testing where the captured microstructure can then be related to the

material’s macroscopic behavior. Kachanov (1958 [68]) introduced the concept

of effective stress theory, which has been successfully implemented to account for the effect of microstructure damage on the mechanical response of a damaged material within the framework of continuum damage mechanics (e.g.

[69], [70], [71], [72], [73], [74], [75]). In continuum damage mechanics,

damage is defined as a microstructural change that induces some deterioration in the material. The effective stress theory postulates that the material damage can be characterized mainly by the decrease in the load-carrying effective area caused by the nucleation and growth of cracks and voids as Murakami (1988

[69]). The theory postulates that a damaged material subjected to a state of

stress can be represented by a perfect material subjected to a fictitious stress called the effective stress. In order to quantify a damage tensor, which is symmetric, the six independent components of the tensor need to be determined

based on measurements of the void fraction in the six directions [76]. The

damage tensor can be incorporated in a continuum model (e.g. [65], [73], [75]).

Chapter 3

Laboratory Testing

Asphalt field cores were tested with SPT in order to measure dynamic modulus and flow number; moreover, an X-ray CT system was used for analysis of aggregate structure and air voids distribution in the asphalt mixture.

3.1 Field Cores

The asphalt cores (100mm diameter x 180mm height) were obtained from a trial field section built by Skanska as part of the construction of a road in

Katrineholm; no traffic loads were applied on the section. The required height for the samples is not typical for Swedish roads and proved a challenge for the laydown and compaction of the mixture. Through experimentation Skanska was able to obtain a single layer of 18cm of HMA from which the samples were cored.

For this project, Skanska used a common base mixture named AG22, which was prepared according to the Trafikverket (Swedish Traffic Administration) specifications; additional information about the asphalt mixture is presented below:

• Binder Type: 70/100

• Binder content: 4.4%

• Air void: 5.6%

• Mixing temperature: 145-155°C [77], [78]

• Compacting temperature: 140-155°C

• Max specific Gravity: 2.512

• Bulk Specific Gravity: 2.372

16 Chapter 3 Laboratory Testing

80

70

60

50

40

30

20

10

0

0

The particles in the distribution of the mixture are shown in Figure 1. One can

see that the mixture meets the gradation requirements according to the Swedish

standard [77].

0.45 Power Gradation Chart

100

90

0,5 1 2 11.2

4 5.6

Sieve size

0.45

[mm]

8

Figure 1. Gradation chart.

Katrineholm AG22 Gradation

Standard AG22 Maxim Gradation

Standard AG22 Minim Gradation

16 22.4

3.2 Testing Equipment

3.2.1 Simple Performance Tester (SPT)

The IPC global SPT (Figure 2) utilized in this work, is a computer controlled

hydraulic loading machine designed to provide researchers and engineers with a tool capable of conducting a range of tests to analyze the performance of HMA.

This device employs hardware technology and software that provides better accuracy, repeatability and operator performance compared to other commercial systems.

The equipment is controlled by a computer system which has installed a

software program (Figure 3) that can be used to perform various tests. The

system gathers the dynamic data from the Linear Variable Differential

Transformer (LVDT) transducers attached to the specimen under test then displays plots appropriate to each test type and the function mode, in real time on the PC.

Testing Equipment 17

Figure 2. The IPC SPT machine.

Figure 3. IPC SPT computer software running dynamic modulus test.

18 Chapter 3 Laboratory Testing

The machine is composed of an electrically powered hydraulic loading system, a confining pressure system, an environmental chamber, and appropriate control systems. For confined tests air pressurization (up to 210kPa) is used as the confining medium. This confining technology is a clean approach for the technician compared to other systems based on oil or water. The test control system is computer based, using sensors on the machine for feedback (load and confining pressure) signals. The hydraulic system uses a bottom loading actuator system with feedback loop control and a run time adaptive control that adjusts the command signal on the fly during testing.

The temperature inside the environmental chamber is changed by a unit outside the triaxial cell controlled by a temperature sensor present inside the chamber.

The machine can change the temperature inside the chamber from negative to positive values (temperatures ranging from -4°C to 60°C) via a small refrigeration unit or a heater unit. Thermally conditioned and pressurized air can be provided to the triaxial cell upon command by the operator, thus providing thermal equilibrium within a three minute time limit.

The studs, where the LVDTs are mounted, can be attached with glue (usually

epoxy) to the samples using a tool called gauge point fixing jig (Figure 4). The

parallel brass studs are glued 100-mm apart and located approximately 25 mm from the top and bottom of the specimen.

Figure 4. Gauge Point Fixing Jig.

The gauges are attached on the sample (Figure 5) in between the studs which

are placed vertically on diametrically opposite sides of the specimen.

Testing Equipment 19

Figure 5. SPT - sample setup.

The LVDTs sensors measure the deformation in sample when a load is applied.

SPT aims to relay asphalt mix design to the performance in the field. Asphalt mixture can be characterized in the laboratory by measuring permanent deformation resistance, fatigue life, tensile strength, stiffness, and moisture susceptibility. Specifically for SPT, the common laboratory test methods for evaluating HMA are: dynamic modulus, flow number (dynamic creep test) and flow time (static creep test).

The tests performed on SPT are detailed below. The procedures for running the

tests to ensure proper testing are presented in Appendix, section A.1.

3.2.1.1 Dynamic Modulus

The dynamic modulus is a relevant property of HMA and has several applications in asphalt pavement technology:

• Visco-Elastic Analysis of asphalt mixtures (laboratory and field)

• Mixture Design and Rutting Resistance: high temperature (fast load rate for freeways, slow loading rate for intersections), plant aged condition, air voids percentage in the mix.

• Mechanistic Empirical Pavement Design (Stiffness, Rutting Model,

Fatigue Cracking Model).

Dynamic Modulus is the ratio of the stress to the strain for asphalt concrete subjected to sinusoidal loading. In the Dynamic Modulus Test while maintaining a specific test temperature the sample is subjected to a controlled sinusoidal (haversine) compressive stress (load) at various frequencies. The applied stress and resulting axial strains are measured as a function of time and

used to calculate the dynamic modulus and phase angle (Figure 6).

20 Chapter 3 Laboratory Testing

Figure 6. Dynamic Modulus - schematic loading.

The dynamic modulus is calculated using the following equation:

E

*

=

σ

ε

o o

φ

=

T i

T p

( 360 )

where:

*

= dynamic modulus

σ

o

= phase angle, degree

= applied stress

ε

o

= measured strain

T

i

= time lag between stress and strain

T

p

= period of applied stress

The dynamic modulus data generated by SPT at different frequencies is organized in the form of arrays, one for time and one for each transducer. The load is measured when applied and the LVDT sensors register the specimen deformation. The analysis has been devised to provide complex modulus in units of Pascals (1 Pa = 1 N/m

2

) and phase angle in units of degrees.

The general approach used here is based upon the least squares fit of a sinusoid,

as described by Chapra and Canale in Numerical Methods for Engineers [79]

and also includes provisions for estimating drift of the sinusoid over time by including another variable in the regression function. The regression approach also lends itself to calculating standard errors and other indicators of data quality.

Testing Equipment 21

The requirements for data quality from the statistical standpoint are given in the

Appendix (Table 8). To meet these requirements one must run several repeated

tests until the results are between the specified limits. In order to cause minimum damage to the samples while measuring the dynamic modulus and phase angle the tests should be conducted by increasing temperature and

decreasing frequency. Two samples were used in each test. The Table 1

summarizes the temperature and frequencies specified for the tests.

Table 1. Characteristics for Dynamic Modulus tests.

Temperatures (°C)

-5

4

20

30

Loading frequencies (Hz)

20,10,1,0.5,0.1

20,10,1,0.5,0.1

20,10,1,0.5,0.1

20,10,1,0.5,0.1

3.2.1.2 Flow Number

Creep is the tendency of a solid material to slowly move or deform permanently under the influence of stresses. The creep curve is created by loading the sample until it fails. The repeated load (dynamic) creep test is used to determine asphalt

permanent deformation parameters and also to estimate or predict rutting [80],

[81]. Tigdemir, M. [82] concluded that repeated loading axial permanent

deformation test can satisfactorily be used for evaluating asphalt concrete mixtures permanent deformation and fatigue characteristics.

Figure 7. Flow Number Test - schematic loading.

22 Chapter 3 Laboratory Testing

The flow number test is a uniaxial repeated load test in which a HMA sample is subjected to cyclic axial load, then the cumulative permanent deformation as a

function of the number of load cycles is measured (Figure 7). Flow Number has

been defined as “The number of load cycles corresponding to the minimum rate of change of permanent axial strain during a repeated load test.”

Results are usually presented in terms of cumulative permanent strain vs. load cycles. The test can be conducted with or without a confining pressure; the dynamic creep test usually better correlates with real field loading conditions

and performance than the static creep test [83].

Fujie, et. al. [84] studied the relationship between the number of load repetitions

and permanent deformation and defined three distinct stages, namely the primary, secondary and tertiary stages.

Figure 8. Cumulative Permanent Strain vs. Load Cycles.

The three major zones (Figure 8) can be detailed as following:

• Primary. Strain rate decreases with loading time.

• Secondary. Strain rate is constant with loading time.

• Tertiary. Strain rate increases with loading time.

Primary stage has high initial level of rutting, with a decreasing rate of plastic deformations, predominantly associated with volumetric change. Secondary stage has small rate of rutting exhibiting a constant rate of change of rutting that

Testing Equipment 23 is also associated with volumetric changes; however, shear deformations increase at increasing rate. While the tertiary stage has a high level of rutting predominantly associated with plastic (shear) deformations under no volume

change conditions [85]. . While the sample remains at relatively constant

volume a large increase in cumulative strain occurs within the tertiary zone.

This large increase is due to shear deformation and the number of load cycles.

The number of cycles at which the sample reaches this large increase - called

flow number (FN) – indicates rutting resistance [3] of a HMA mixture. Larger

permanent deformation in the field is inverse proportional to the value of flow numbers (the lower the flow numbers the higher the deformation in the field).

Because of this correlation a minimum acceptable flow number requirement can be established for asphalt mixtures.

The secondary zone appears in the linear portion of the cumulative strain curve which is modeled by an equation of the form: 𝜀 𝑝

= 𝑎𝑁 𝑏 where:

ε p

N

a

b

=

=

=

= cumulative permanent strain number of loading cycles y-intercept of total cumulative strain curve slope of total cumulative strain curve

The values of “a” and “b” are usually calculated and reported for each mixture.

As mentioned before, the flow number is the number of test cycles required until tertiary flow starts in the mixture. The higher the flow number, the longer the time until the tertiary flow in the mixture stars. The flow number varies with

the change in the asphalt content and percentage of air voids in the HMA [86].

Rutting has been considered the most serious distress in flexible pavement and is caused by the accumulation of the permanent deformation (NCHRP 9-33-

Tentative criteria, “Mix design manual for hot mix asphalt”) [9].

The flow number test is based on the result from repeated loading and unloading of HMA sample and the deformation of the specimen is recorded as a

function of load cycles [87]. The Simple Performance Test machine records the

strain during the repeated loading. For 0.1 seconds a load is applied to the sample and then is followed by 0.9 seconds of rest time (dwells) is applied to

the specimen [13], [88].

The flow number test will be performed at a high pavement temperature representative of the project location and pavement layer depth to evaluate the rutting resistance of the mixture. For the specific mixture used in this thesis project (AG22), the Swedish standard recommends a testing temperature of no

more than 40°C [89]. Figure 9 illustrate Flow number loading.

24 Chapter 3 Laboratory Testing

CYCLE 1 CYCLE 2

δ

P

(1)

CONTACT DEVIATOR STRESS +/- 2%

REPEATED DEVIATOR STRESS +/- 2%

CONFINING PRESSURE +/- 2%

δ

P

(2)

0.1

0.9

TIME, SEC

Figure 9. Repeated Load Test principle - schematic of flow number test loading.

A major assumption in the flow number test is that the stresses are distributed uniformly over the specimen. Friction between the loading platen and the specimen produces shear stresses which result in a deviation from this assumption. The effects of friction can be minimized by using long specimens.

The test specimen size for the simple performance tests was determined in an extensive specimen size and geometry study conducted in Project 9-19. The specimen diameter of 100 mm was selected to provide flow data that are independent of specimen size. The height to diameter ratio of 1.5 was selected to provide dynamic modulus and flow data that are independent of specimen height. The reduction of end friction in these tests was a significant factor in the recommendation for specimen size.

3.2.2 X-Ray Computed Tomography (CT) System

An X-View™ X5000-CT Computed Tomography System (Figure 10), owned

by the Royal Institute of Technology KTH, was utilized during this project. The system is a seven-axis universal X-ray imaging system designed for the inspection of large objects with a flat panel digital plate. The 5000 Series has an innovative top load cabinet design for easy part loading. It can accommodate a

Testing Equipment 25 variety of part shapes, sizes and weights and its scanning X-ray energy range intensity can be selected from two energy sources: 225kV and 450kV.

Figure 10. The CT machine X-5000.

26 Chapter 3 Laboratory Testing

When X-ray penetrates into the asphalt mixture, the ray intensity becomes attenuated due to the absorption of atoms in the material. The grey levels in a

CT slice image correspond to X-ray attenuation which is the proportion of

X-rays scattered or absorbed as they pass through the sample. Different

materials attenuate X-ray at different rate (Figure 11); materials with higher

density have larger attenuation coefficients. In order to determine the internal structure of the specimen, one should calculate the attenuation coefficient via a process of computerized tomography.

Linear Attenuation Coefficients for different materials

10

5

Aggregate

Bitumen

Air

10

0

10

-5

10

-3

10

-2

10

-1

Photon Energy in MeV

10

0

10

1

Figure 11. Attenuation coefficients for different materials [90].

The main components of an X-ray tomography image system are:

• X-ray sources

• A series of detectors that measure X-ray intensity along multiple beam paths (linear or planar detector)

• A rotational specimen manipulator

• A collimator (used for linear detector array)

Figure 12. General mechanism of X-ray tomography scanning.

Testing Equipment 27

In the simplest approach, the source generates X-Ray radiation with certain intensity that passes through the specimen along different paths in several directions and a set of CT images is produced. The intensity of the X-Rays is measured before they enter the specimen and after they penetrate through it.

The intensities of the transmitted X-Rays are recorded on the detectors placed at the other side of the specimen. The scanning of a slice is completed after collecting the intensity measurements for a full rotation of the specimen. The specimen is then shifted vertically by a fixed amount (the slice thickness) and the entire procedure is repeated to generate additional slices.

The intensity values are used to calculate the distribution of the linear attenuation coefficient within a specimen. The resulting X-Ray CT image is a map of the spatial distribution of the linear attenuation coefficient. In this map, brighter regions correspond to higher values of the coefficient. Higher values of the attenuation coefficient correspond to regions with higher density. Therefore, since the linear attenuation coefficient at each point depends directly on the density of the specimen at that point it is feasible to distinguish the different features of HMA.

As mentioned before, the ability of the X-rays to differentiate materials depends

on their respective linear attenuation coefficients [91]. Materials with mass

attenuation coefficient can be obtained to determine the energy level that is most appropriate for the asphalt concrete sample scanning.

The X-ray scanning process includes warm-up the system, scan the specimen, calibrate the system, and reconstruct the element. An industrial computed tomography software called efX CT was used for visualization, calibration and reconstruction. After reconstruction, Avizo Fire, a 3D Analysis Software for

Materials Science was used for obtaining and visualizing advanced qualitative and quantitative information on material structure images.

3.2.2.1 Warm up the system

Run the fxe-control application to start the warming process; place the beam blocker in front of the X-ray source to protect the detector from the unfiltered

X-rays generated during the warming up procedure.

3.2.2.2 Scan the specimen

This step should not be rushed as the quality of the CT depends on the quality of the acquisition. Ensure that there are no saturations in any area of interest at every degree of rotation, by selecting an appropriate voltage and current configuration. Additionally, try to have the gray level values of all areas of interest in the area of 15 to 75 percent of the total available gray levels of the operating bit depth.

28 Chapter 3 Laboratory Testing

The quantities of images used for the scan vary on part geometry (inside and out) and resolution sought. Use more images for higher detail and complex parts. Rectangular parts, for example, where the gray level values are border line when imaged through the thickest region, should have more images used during acquisition so that there will be more data for reconstruction, from the images of the thinner region. The typical number of images captured for CT scan is 360, 720, and 1440.

Figure 13. X-ray CT - sample setup.

The number of frames per second (fps) used are directly related to how much detail one wants to capture during the scan. A number of 2fps is usually ideal for small objects, but a greater value is recommended objects like asphalt cores.

There is a tradeoff between scanning-speed and scan quality when considering the number of frames per second.

The process of re-alignment of the detector, X-ray source and the scanning platform can be monitored using the CCTV monitor outside the scanning

chamber; there are four cameras installed in the scanning chamber (Figure 13).

3.2.2.3 Calibrate the system

Calibration routine includes two steps: capturing images of a calibration block and capturing images of the background.

Calibration Block Radiographs

The calibration block radiographs are used to create an accurate 3D rendering of the part. It also establishes the relationship between voxels and units of

Testing Equipment 29 measurement enabling the user to take accurate measurements from the volume data. The calibration block shall be imaged at the exact same geometry settings as the part. The X-ray tube energies can be varied though. In fact, they should be varied to maximize the contrast sensitivity between the calibration spheres and the surrounding material. The typical number of images used for this step is

60 [91].

It is imperative that no movements be made with the exception of the rotational axis. If the calibrations tool needs elevation it is recommended that it is place on a stable object. Capturing of these images can be done before or after capturing of the inspection radiographs. Additionally, the energies and filters used during acquisition may be altered for the calibration images. The idea is to maximize the contrast between the calibration spheres and the background. Several tests were run to be able to set the required filters for scanning the samples.

Figure 14. The calibration rod.

For this work, the 15mm calibration rod was selected; the Figure 15 shows the

calibration process by using efX CT software. Positioning the calibration tool will require a degree of trial and error. The ideal situation will be to have the calibration tool travel from edge to edge of the image during a single 360 degree of rotation. The balls of the calibration tool must be at least 1 ball diameter from the edge of the image. It is also beneficial but not necessary to have the calibration tool reach from the bottom of the detector to the top. If the tool does not span from the top to the bottom, place the tool so that one portion of it reaches the top or bottom. This allows the creation of larger ellipses during rotation and better data for the software to build the volume off.

30 Chapter 3 Laboratory Testing

Figure 15. Images of the calibration tool.

The accuracy of the software in interpreting the geometric positioning of the

X-ray tube, detector, and manipulator is given by the number of calibration spheres that are visible.

Background Radiograph

The background radiograph is an optional step that can improve the quality of the 3D rendering. The background image is captured at the same geometry settings and energies as the part even if the background is being saturated during acquisition. This image is used by the software as a means to improve quality by subtracting that image from every part radiograph.

Artifacts such as beam hardening, ring artifacts, etc. generated by defective pixels which affect the quality of the acquired scanned image may be corrected during the detector calibration stage.

3.2.2.4 Reconstruction

Reconstruction is done via mathematical process that converts the raw data into image slices. During this process the intensity data in the sinogram are mapped to CT values that have a range determined by the computer system (16 bit, 32 bit, 64 bit, etc.). For most industrial scanners, these values map to the grayscale in the image files produced by the systems.

Testing Equipment 31

The size of the reconstruction matrix is determined by the number of views and the number of measurements per view. Spatial resolution in an image can be improved by reducing the pixel size. However, after a certain limit smaller pixels do not increase the spatial resolution anymore and can induce artifacts in the image. Reconstructing with smaller pixels, under certain circumstances can be a useful technique.

After scanning the machine is shut down and the reconstruction phase begins.

The reconstruction program runs for around half an hour and constructs the image of the sample. The image is then further adjusted by using a histogram to achieve the best contrast and visibility of the sample mixture.

3.2.2.5 Post-processing

The Avizo Fire software package was used to perform analysis on the X-ray CT images. Avizo Fire has a broad range of software tools for obtaining and visualizing advanced qualitative and quantitative information on material structure images. The following techniques were applied to analyze aggregate structure and air void distribution:

• Data import from CT-scans

• Scaling, calibration, conversion, re-sampling

• Image enhancement, comprehensive filtering and convolution

• Thresholding and auto-segmentation, object separation, automatic labeling

• Skeletonization

• Direct volume visualization

• Automatic or interactive segmentation

• 3D geometry reconstruction

• Orthogonal, oblique, cylindrical, and curved slicing

• Quantification and analysis

• Results viewer with spreadsheet tool and charting

• Automatic individual feature measurements, 3D localization, and spreadsheet selection

• Automated statistics, distributions graphs

Chapter 4

Test Results

4.1 Field Core Samples

A total of twelve specimens were tested in this project; their correspondent identifications (IDs), tests performed and specific test objective are presented in

Table 2.

Table 2. The samples.

FN4

CT1

CT2

CT3

CT4

Sample Code Test performed

T1 Dynamic modulus

T2

D1

Dynamic modulus

Dynamic modulus

D2

FN1

FN2

FN3

Dynamic modulus

Flow Number

Flow Number

Flow Number

Flow Number

Loading 600 Cycles

Loading 600 Cycles

Loading 1200 Cycles

Loading 1200 Cycles

Analysis

Effects of sample dimensions

Effects of sample dimensions

Dynamic modulus Master curve

Dynamic modulus Master curve

Microstrain and failure point

Microstrain and failure point

Microstrain and failure point

Microstrain and failure point

X-ray CT before and after loading

X-ray CT before and after loading

X-ray CT before and after loading

X-ray CT before and after loading

34 Chapter 4 Test Results

4.2 Simple Performance Tester Results

This section presents and discusses the results obtained from SPT. All tests were performed on unconfined samples. The dynamic modulus test was measured on samples T1, T2 to study the effects of the sample geometry. The dynamic modulus master curve was created using samples DM1 and DM2 to analyze the HMA. The flow number test was performed in samples FN1 to FN4 to record the microstrain accumulation during loading; Flow Number was afterwards used to calculate the range of the secondary stage of dynamic creep curve so that two interest loading points could be selected for further sample analysis using X-ray CT.

4.2.1 Dynamic Modulus Test

The SPT software reports the average dynamic modulus for the specimen at each temperature and frequency tested. At lower temperatures, when stiffness is higher the load had to be increased from 0.06 to 6 kN. Several tests were run to find an adequate load to use. The contact stress was adjusted automatically when the load was changed.

6000

5000

5325

5029

4000

4019

3733

3000

2930

2000

1000

0

0

1475

1069

738,7

5 10 15

Frequency (Hz)

20

Third measurement

Second measurement

First Measurement

25 30

Figure 16. Dynamic Modulus - sample rotation.

Simple Performance Tester Results 35

To analyze the effects of the sample geometry (irregularities in the specimen induced during coring) several tests were performed with the sample rotated vertically, switching transducers (rotating sample 120°). A statistical analysis

(regression) can be found in Table 9, section A.2 of Appendix; Dynamic

modulus test was performed at 20°C three times by rotating sample T1 120° for

each test to see if there are large differences in the results (Figure 16).

Dynamic modulus at 20°C was also measured applying different loads (2KN and 6KN) on sample T2; also dynamic modulus was obtained when the same sample was deliberately misaligned (not centered) during testing; the results are

presented in Figure 17. The variation of dynamic modulus in these cases is

smaller than 10%, e.g. meeting SPT standards [4].

6000

5000

5350

4642

4069

4000

3147

3488

3000

2631

2000

Specimen loaded- 6kN

1000 Specimen loaded- 2 kN

649

Misaligned specimen

0

0 5 10 15

Frequency (Hz)

20 25 30

Figure 17. Dynamic Modulus - different loads and misalignment.

During the dynamic modulus test several temperatures and load frequencies were set according to the actual road environment. For Sweden weather conditions, dynamic modulus should be measured at low temperature. Hirsch

model [92], (see also Appendix, section 0) was developed to estimate the

dynamic modulus at low temperatures from the master curve calculated at higher temperatures.

In general, to generate the dynamic modulus master curve, several steps must be followed:

• Measure dynamic modulus for minimum two samples.

• Plot the measured dynamic modulus values on a logarithmic scale.

36 Chapter 4 Test Results

• Choose a reference temperature.

• Use superposition principle and calculate shift coefficients.

• Generate fitting curve according to the shift coefficients found.

The procedure followed to calculate the dynamic modulus master curve for the

asphalt mixtures evaluated in this project is showed in Figure 18 and Figure 19.

As expected, one can see that dynamic modulus increases when the loading frequency increases; also dynamic modulus decreases with the increase in temperature.

3500

3000

2500

2000

1500

1000

500

0

0 5 10 20 25

-5°C

1°C

10°C

20°C

35°C

30

Figure 18. Dynamic Modulus [ksi] vs. Frequency [Hz].

10000

1000

100

10

-5

°C

1

°C

10

°C

20

°C

35

°C

1

0,1 1

Frequency (Hz)

10 100

Figure 19. Dynamic Modulus [ksi] vs. Frequency [Hz] - logarithmic scale.

Simple Performance Tester Results 37

From the results at different temperatures the master curve is created by using the superposition principle and choosing a reference temperature; in this thesis

20°C was selected.

As part of a research project for evolution SPT, an Excel workbook capable to solve the specific modified version of the Mechanistic-Empirical Pavement

Design Guide master curve equation was developed [93]. This Excel workbook

containing the Mastersolver developed by NCHRP to obtain the Witczak (Shift) coefficients and the master curve was used (All the calculations for the master

curve can be found in Appendix, Table 10). The obtained dynamic modulus

master curve is presented in Figure 20 and Figure 39.

10000

1000

100

-5°C

1°C

10°C

Fit

20°C

35°C

10

1

1,E-03 1,E-01 1,E+01 1,E+03

Reduced Frequency, Hz

Figure 20. Dynamic Modulus [ksi] vs. Reduced Frequency [Hz].

1,E+05

45

40

35

30

25

20

-5°C

1°C

10°C

20°C

35°C

15

10

5

0

1,E-06 1,E-04 1,E-02 1,E+00 1,E+02 1,E+04 1,E+06

Reduced Frequency, Hz

Figure 21. Phase Angle [deg] vs. Reduced Frequency [Hz].

38 Chapter 4 Test Results

The phase angle measures how quickly HMA can recover from strain. A phase angle of 0 degrees is for elastic materials and one of 90 degrees corresponds to viscous materials; consequently HMA falls in between as a viscous-elastic

material. The phase angle at reduced frequency is presented in Figure 21. The

results are typical for HMA where the phase angles decreases with lower temperature and increases with higher temperature. For high temperatures the phase angle increases when the frequency is increased because the sample is more elastic than at low frequencies.

The Federal Highway Administration from USA in cooperation with asphalt researchers and manufacturers developed knowledge containing typical dynamic modulus results for several frequently used mixtures. In Sweden such information is not yet available; although some projects were started to develop

it. One Swedish project [94] aimed to verify if the dynamic modulus results

produced by different laboratories, using the same asphalt mixture are comparable; they found significant variability for the laboratories evaluated.

The variability for dynamic modules may have several reasons: different machines, different environmental condition, different loads, different testing procedures, etc; these conclusions highlighted how complex and sensitive dynamic modulus test on asphalt mixtures can be.

As a reference; the dynamic modulus master curve calculated from this project, was compared with test results produced in other two studies: Richard Nilsson’s

project sponsored by SBUF [95], and Eric Oscarsson’s Licentiate Thesis, Lund

University [96]. The comparison is presented in Figure 22 and Figure 23.

10000

1000

100

10

1

1,E-03 1,E-01 1,E+01 1,E+03

Reduced Frequency, Hz

Fit curve

Fit curve Oscarsson

Fit curve Nilsson

1,E+05

Figure 22. Master Curve comparison (Test results, Oscarsson and Nilsson).

Simple Performance Tester Results 39

This thesis test results at 35°C and the results from Nilsson (at 30°C) are rather similar with small differences at lower frequencies and high temperature. They might also have used a lower load, or because the dynamic modulus tests were unconfined (Nilsson’s curve is similar when the difference between confined

and unconfined tests is taken into account as underlined in NCHRP 547 [97]).

There are large differences between this thesis test results and the results of

Oscarsson. These differences can have several reasons: not enough data detail available, different testing machine, different air voids, different compaction methods, different temperatures, etc. Similar conclusions can be drawn from

phase angle comparison, as can be seen in Figure 23.

50

45

40

35

30

25

20

15

Results

10

Oscarsson Results

5

Nilsson Results

0

1,E-06 1,E-04 1,E-02 1,E+00 1,E+02 1,E+04 1,E+06

Reduced Frequency, Hz

Figure 23. Phase Angle comparison- (Test results, Oscarsson and Nilsson).

Additionally, statistical analyses were performed; Figure 24 and Figure 25 show

a statistical comparison (regression analysis for the test of effects between subjects) for a reference temperature (20°C) calculated with the IMB SPSS statistic suite version 19.

The statistical analysis provides a better visualization of the differences between thesis test results for dynamic modulus and the data from Nilsson and

Oscarsson. In Figure 24 an estimation of 95% confidence interval was used for

the three data sets: thesis data, Nilsson, Oscarsson. One can see that the median for thesis data is very close to the median for Nilsson, while as expected from previous discussion the mean from Oscarsson is not similar.

Another statistic analysis based on stem-and-leaf comparison is presented in

Figure 25 for different frequencies. All results have the same trend and thesis

40 Chapter 4 Test Results

data and Nilsson’s are close together. One can conclude that the results in the thesis are similar enough to other field data, considering the complexity and variability of dynamic modulus test.

Figure 24. Comparison of Dynamic Modulus for different labs at a reference temperature of 20°C.

Figure 25. Data comparison - stem-and-leaf.

Simple Performance Tester Results 41

4.2.2 Flow Number Tests

The resistance of asphalt mixtures to permanent deformation is measured by flow number (dynamic creep test). In response to creep loading, both static and cyclic, asphalt concrete materials develop permanent deformation which accumulates with time or number of load repetitions. This accumulated permanent deformation is the cause of rutting in asphalt pavements. In this thesis the flow number (dynamic creep tests) was performed to determine after how many load cycles the typical sample used in this thesis is damaged

(cracked).

According to the Swedish test recommendations the creep test should be conducted at maximum 40°C for this specific mixture (AG22); for this work, all

the flow number tests were conducted at this temperature [77].

Flow number test results for each sample are summarized in Table 3.

Table 3. Flow Number - test results.

Sample Code Micro(µ)Strain Loading Cycles Minimum Strain Rate

FN1

FN2

34067

35035

1498

1505

20.3

21.4

FN3 24077 1390 18.2

FN4 30641 1520 16

Flow number was measured for five samples (FN1 to FN4 and DM1); these

results are plotted in Figure 26. Sample DM1 was used before for dynamic

modulus measurements that means that the sample has been previously subjected to different temperatures and small loads which may explain the its low strain; it is recommended to use new samples for different tests.

The average failure point for the asphalt mixture was estimated to be around

1400 load cycles, according to the Flow Number test performed according to the NCHRP specification. An example of the results of the Flow Number test

for sample FN2 can be seen in Figure 27.

In order to analyze the aggregate structure and air voids distribution of the specimens with the X-ray CT system, two reference loading points were selected; these loading points were specifically chosen such that the sample is not in the failing stage (tertiary stage of the creep curve).

42 Chapter 4 Test Results

50000

45000

40000

35000

30000

25000

20000

15000

10000

5000

0

0

Sample FN2

Sample FN1

Sample FN4

Sample FN3

Sample DM1

35035

11836

24077

34067

30641

50000

45000

40000

35000

30000

25000

20000

15000

10000

5000

0

0

500 1000 1500

Cycles

2000

Figure 26. Flow Number for different samples.

2500

Accumulated Permanent Strain- Sample FN1

Flow Number

Strain Rate

250

200

Flow Number = 1 505

Minimum strain rate = Flow number

150

100

50

500 1000

Cycle Number

1500

Figure 27. Flow Number Test Result for Sample FN2.

2000

0

3000

Simple Performance Tester Results 43

Knowing the approximate number of the cycles when the sample breaks

(approximately 1400 load cycles), the reference point 1 (Figure 28) was

selected to be close to the limit between primary and secondary stages (around

600 load cycles); while the reference point 2 was intended to be close to the limit between the secondary and tertiary stages, without reaching failure

(around 1200 load cycles) [2].

Figure 28 Creep curve - test stages.

The dynamic modulus results for all CT samples before and after loading, as

well as number of load cycles and microstrain are presented in Table 4 and

Figure 29.

One can see from Table 4 and Figure 29 that there are big differences between

sample sets:

• CT1 and CT2 have similar dynamic modulus and big difference in microstrain accumulation. Interestingly enough,

• CT3 and CT4 have similar dynamic modulus (Lower that CT1 and CT2) and big difference in microstrain accumulation.

• CT1 and CT3 have similar microstrain accumulation, knowing that CT3 was subjected to higher number of load cycles. CT2 and CT4 show similar trend.

• CT1 and CT2 have normal behavior in dynamic modulus as it decreases after loading but, unexpectedly, for CT3 and CT4 the dynamic modulus increases after loading. This behavior could eventually be explained by higher segregation in the CT3 and CT4 set. From visual inspection, it was concluded that samples CT1 & CT2 looked coarser and had smoother external surfaces, than samples CT3 & CT4.

44 Chapter 4 Test Results

18000

16000

14000

12000

10000

8000

6000

4000

2000

0

0 1000

Sample CT2

Sample CT1

Sample CT4

Sample CT3

1200 1400 200 400 600

Cycles

800

Figure 29. Microstrain Samples CT1 to CT4.

Table 4. Samples CT1-CT4 – Dynamic Modulus and Load Cycles.

Sample

Code

Load

Cycles

Dynamic modulus (MPa) before loading after loading

MicroStrain accumulation

CT1 600 6195 5582 12159

CT2 600 6109 4726 16114

CT3 1200 4951 6204 10645

CT4 1200 4626 6062 15705

The main finding from SPT was that significant variability in dynamic modulus

(before and after loading) and microstrain accumulation from flow number test was identified; most likely, induced by irregularities in the sample geometry

(wall surface) and segregation in the field cores.

X-ray Computed Tomography Results 45

4.3 X-ray Computed Tomography Results

As mentioned before, the main objective of using X-ray CT was to evaluate aggregate structure and air void distribution for each sample before and after

SPT loading; voxel size and resolution for each reconstructed volume are

presented in Table 5.

Table 5. Volume resolution.

Sample Code Voxel size (microns) Resolution (voxels)

CT1 Before Loading 91.000 x 91.000 x 91.000 1453x1843x1414

CT1 After Loading 93.000 x 93.000 x 93.000

CT2 Before Loading 91.000 x 91.000 x 91.000

CT2 After Loading 93.000 x 93.000 x 93.000

1297x1778x1314

1401x1846x1446

1348x1838x1319

CT3 Before Loading 105.000 x 105.000 x 105.000 1286x1664x1334

CT3 After Loading 105.000 x 105.000 x 105.000 1185x1660x1216

CT4 Before Loading 105.000 x 105.000 x 105.000 1303x1962x1338

CT4 After Loading 105.000 x 105.000 x 105.000 1274x1722x1286

During the Post-processing stage, beam hardening which is the most commonly encountered artifact in CT scanning was detected, regardless the use of 5mm cupper filtering plates during testing. Beam hardening causes the edges of an object to appear brighter than the center, even if the material is the same

[98] throughout. The artifact derives its name from its underlying cause: the

increase in mean X-ray energy or "hardening" of the X-ray beam as it passes through the scanned object. Because lower-energy X-rays are attenuated more readily than higher-energy X-rays, a polychromatic beam passing through an object preferentially loses the lower-energy parts of its spectrum. The end result is a beam that, though diminished in overall intensity, has a higher average energy than the incident beam.

In X-ray CT images of sufficiently attenuating material, this process generally manifests itself as an artificial darkening at the center of long ray paths, and a

46 Chapter 4 Test Results

corresponding brightening near the edges. In objects with roughly circular cross sections this process can cause the edge to appear brighter than the interior.

Beam hardening can be a pernicious artifact because it changes the grey level of a material depending upon its location in an image. Thus, the attempt to utilize a single CT number range to identify and quantify the extent of a particular material can become problematic. One measure that is sometimes taken is to remove the outer edges of the image and analyze only the center. Although this technique removes the worst part of the problem, the artifact is continuous and thus even subsets of the image are affected. Furthermore, if the cross-sectional area of the object changes from slice to slice, the extent of the beam-hardening artifact also changes, making such a strategy prone to error.

In this thesis work, when beam hardening was detected, it was decided to crop

1.5 cm form the top and from the bottom of each sample in order to minimize the beam hardening effect.

4.3.1 Air Void Distribution

It is known that the air void distribution strongly influences the mechanical response of HMA mixtures

[99], [100].

Considering that the samples for this study are field cores; then uneven air void distribution induced during the construction process may be expected

[101].

Nilsson reported that a reduction in air void content caused an increase in asphalt mix stiffness. Air voids in the center of a field core may be significantly lower than those of the entire cylinder (about 0.2% to 1.5%, depending on the type of mix)

[102] [103].

The distribution of air voids in the sample CT1 along the vertical direction,

before and after loading, is illustrated in Figure 30. The air voids distribution is

relatively homogeneous between 0-40mm (top), has a high and low peak between 40mm and 60mm (middle) and has a wide increase between 60mm and

120mm (bottom). After loading, the overall air void content of the sample slightly decreased although locally it varies from low at the top to larger at the bottom.

The concentration of air voids before and after loading for sample CT2 is given

in Figure 31. The air void of CT2 has a different distribution compared to

sample CT1 and varies widely between 8% and 16%. This non-homogenous air void distribution might be caused by irregularities during the construction process. The trial error process during construction to obtain an 18 cm thick asphalt layer seems to be the main factor for the significant variation of air void content among samples. Furthermore, air void distribution before and after loading does not change; this implies that no re-accommodation of particles may be attributed to the applied load.

X-ray Computed Tomography Results 47

0,1

0,09

0,08

0,07

0,06

0,05

0,04

0,03

0,02

0,01

0

0 20 40 60 80

Depth (mm)

Sample CT1 Before Loading

Sample CT1 After Loading

100 120 140

Figure 30. Air voids distribution for sample CT1 before and after loading.

0,18

0,16

0,14

0,12

0,1

0,08

0,06

0,04

0,02

0

0 20 40

Sample CT2 Before Loading

Sample CT2 After Loading

60 80

Depth (mm)

100 120 140

Figure 31. Air voids distribution for sample CT2 before and after loading.

48 Chapter 4 Test Results

0,12

0,1

0,08

0,06

0,04

0,02

Sample CT3 Before Loading

Sample CT3 After Loading

0

0 20 40 80 100 120 140

Figure 32. Air voids distribution for sample CT3 before and after loading.

0,12

0,1

0,08

0,06

0,04

0,02

0

0 20 40

Sample CT4 Before Loading

Sample CT4 After Loading

60 80

Depth (mm)

100 120 140

Figure 33. Air voids distribution for sample CT4 before and after loading.

X-ray Computed Tomography Results 49

Sample CT3 air voids distribution, presented in Figure 32 has an increasing air

void content from top to bottom with almost double content at the bottom. After loading the air void content at the top decreased while on the bottom increased; a small overall increase in air voids through the whole sample was loadinduced.

The air voids distribution for sample CT4 shown in Figure 33 exhibited higher

air void content at top and bottom; also, an increase of air voids content after loading can be observed.

The average air voids content for all the

samples is presented in Table 6. One can

see that the air void content of sample CT1, which strain level, is close to one percent, decreases after loading. This could be explained according to Tashman

et al (2004 [104]); they reported that “At initial stage of deformation, the air

voids tended to contract and the existing microcracks tended to close up as indicated by the negative percent change in the void content at one percent strain” during triaxial compression tests of asphalt mixtures at high temperatures.

Table 6. Total air voids content for samples CT1-CT4 before and after loading.

Sample Code

Before loading (%)

CT1

4.7

CT2

10.5

CT3

4.7

CT4

5.3

After loading (%) 4.2 10.5 5 5.9

For sample CT2 the air void content before and after loading stayed constant; this behavior appears due to variation in air void content with the sample height and because some of the air voids compact and some increase as presented in

Figure 31.

Samples CT3 and CT4 (loaded with 1200 cycles) present an increase in air void content, probably induced by the movement of aggregates under higher cumulative load.

One of the options available in Avizo Fire suite is to measure and classify the dimension of the air voids. The air voids histogram, before and after loading

sample CT1, is shown in Figure 34 (top). After loading sample CT1 the air

voids from class [7535-75357] decrease and migrate to all other classes but mostly to class [735-7535] which grew more than twice. This suggests that densification of the material due to re-accommodation of particles has occurred.

Air voids dimension classification histogram before and after loading for

sample CT4 is presented in Figure 34 (bottom). Interestingly enough, one can

also see that sample CT4 has no middle size air voids.

50 Chapter 4 Test Results

0.012

0.01

0.008

0.006

0.004

0.002

0

0.02

0.018

0.016

0.014

Sample CT1 Before loading

Sample CT1 After loading

0- 0.75

0.75- 7.5

7.5- 75 75- 753

Size void classification (mm

3

)

753- 7535 7535- 75357

0.05

0.045

0.04

0.035

0.03

0.025

0.02

0.015

0.01

0.005

Sample CT4 Before loading

Sample CT4 After loading

0

0- 0.75

0.75- 7.5

7.5- 75 75- 753

Size void classification (mm

3

)

753- 7535 7535- 75357

Figure 34. Air voids size histogram before and after loading for samples

CT1 and CT4.

X-ray Computed Tomography Results 51

0.05

0.04

0.03

0.02

0.01

0

0.1

0.09

0.08

0.07

0.06

Sample CT2 Before loading

Sample CT2 After loading

0- 0.75

0.75- 7.5

7.5- 75 75- 753

Size void classification (mm

3

)

753- 7535 7535- 75357

0.02

0.018

0.016

0.014

Sample CT3 Before loading

Sample CT3 After loading

0.012

0.01

0.008

0.006

0.004

0.002

0

0- 0.75

0.75- 7.5

7.5- 75 75- 753

Size void classification (mm

3

)

753- 7535 7535- 75357

Figure 35. Air voids size histogram before and after loading for samples

CT2 and CT3.

52 Chapter 4 Test Results

The air voids histograms for CT2 (top) and CT3 (bottom) can be observed in

Figure 35. There is a tendency for smaller air voids classes to migrate to larger

ones in all the samples except CT1.

4.3.1 Aggregate Structure

In order to verify if segregation played a role in the variation of SPT results, the

X-ray CT system was utilized to analyze the field cores.

As a reference, Figure 36 shows raw images corresponding to CT1 and CT4

samples; it can be noticed that CT1 has less air voids, bigger particles, and better aggregate interlock than CT4. In addition, segregation was detected in sample CT4 (fine aggregates localized at the bottom) and concentration of bigger air voids in the middle. Sample CT4 looks like a two-layer system; having a coarse layer on the top and a fine-medium layer at the bottom.

a) Sample CT1. b) Sample CT4.

Figure 36. X-ray CT image.

X-ray Computed Tomography Results 53

A 3D reconstruction of the sample CT1 was done by using Avizo Fire; some

beam hardening was detected (Figure 36) and consequently the volume had to

be cropped. Avizo Fire suite was also used to segment the aggregate particles

and create the shown 3D cube for further analysis (Figure 37).

Figure 37. Distribution of aggregates in a 3D volume of interest of sample CT1.

Finally, the corresponding particle size distribution was calculated for the 3D

volume of interest (Figure 38) to better analyze the internal structure of the

54 Chapter 4 Test Results

asphalt mixture. The 3D volume of interest presents uneven sizes of aggregates: a larger number of small-to-medium size than medium-to-large size ones.

0.2

0.15

0.1

0.05

0.35

0.3

0.25

0.5

0.45

0.4

0

1- 1000

Gradation Example

1000- 2000 2000- 3000

Size aggregates classification (mm

3

)

3000- 4000 4000- 5000

Figure 38. Example of aggregate size classification in the 3D volume of interest.

The application of X-ray CT and complex 3D image analysis are very powerful applications and it could eventually provide a non-destructive tool for verification of gradation of asphalt mixtures.

Chapter 5

Closure

5.1 Conclusions

SPT results from asphalt field cores, including dynamic modulus (before and after loading) and microstrain accumulation (flow number), exhibited significant variability; most likely, induced by irregularities in the core shape.

The analysis of aggregate structure and air voids distribution performed trough

X-ray CT, clearly identified segregation in the asphalt mixture as a another key factor that induced variability in SPT results.

SPT Dynamic modulus values measured during this work are comparable with data obtained from other research projects that investigated typical Swedish asphalt mixtures.

Significant variability in flow number results was found; in consequence the number of cycles to failure was not enough to induce failure of the sample. Not even visible damage was induced under the applied number of load repetitions.

Beam hardening was detected during the reconstruction process, regardless the use of 5mm cupper filtering plates during scanning; this increased the difficulty of key post-processing tasks including quantification of air voids and segmentation of aggregate particles.

The X-ray CT technology provides valuable information about the internal structure of asphalt mixtures, including aggregate particle size distribution and analysis of air voids structure (size, distribution and connectivity). In summary,

X-ray CT is a very powerful and promising technology that allows enhancing understanding of asphalt mixture failure mechanisms, which may eventually generate further development of asphalt mixture design procedures and/or optimization of pavement construction methods.

56 Chapter 5 Closure

5.2 Recommendations

Research should continue to further develop and refine the promising X-ray CT analysis; this technique may generate fundamental knowledge about the effects of internal structure of asphalt mixtures on asphalt field performance; this could eventually lead to the design and construction of rutting and cracking resistant asphalt mixtures.

Specifically, the following areas also need further development:

• X-ray CT scanning technique.

• X-ray CT reconstruction software

• X-ray CT post-processing

Bibliography

1. NCHRP-Project_9-19, Witczak, M.: Superpave support and performance models management., University of Maryland (1999-2006)

2. NCHRP-Project_9-29, Bonaquist, R.: Simple Performance Tester for

Superpave Mix Design. (2001- 2011)

3. NCHRP-Report_465, Witczak: Simple Performance Test for Superpave Mix

Design., Washington D.C. (2002)

4. NCHRP-Report_513, Bonaquist, R. F., Christensen, D. W., Stump, W.:

Simple performance tester for Superpave Mix Design: First-Article

Development and Evolution., Washington (2003)

5. NCHRP-Report_530, Christensen, D. W., Bonaquist, R. F.: Evaluation of

Indirect Tensile Test (IDT) Procedures for Low-Temperature Performance of Hot Mix Asphalt., Washington (2004)

6. NCHRP-Report_614, Bonaquist, R.: Refining the Simple Performance

Tester for Use in Rutine Practice., Washington (2008)

7. NCHRP-Report_629, Bonaquist, R.: Ruggedness Testing of Dynamic

Modulus and Flow Number Tests with the Simple Perfprmance Tester.,

Washington (2009)

8. NCHRP-Report_702, Bonaquist, R.: Precision of the Dynamic Modulus and

Flow Number Tests Conducted with the Asphalt Mixture Performance

Tester., Washington (2011)

58

9. NCHRP-Project_9-33, Christensen, D.: A Mix Design Manual for Hot Mix

Asphalt. (2005- 2010)

10. Denison, C., Carlson, W. D., Ketcham, R. A.: Three-Dimensional

Quantitative Textural Analysis of Metamorphic Rocks Using High-

Resolution Computed X-Ray Tomography: Part I. Methods and

Techniques. Journal of Metamorphic Geology 15, 29- 44 (1997)

11. Denis, M. J.: Industrial Computed Tomography. NDE Systems and Services,

General Electric Company (1997)

12. Masad, E., Muhunthan, B., Shashidhar, N., Harman, T.: Internal structure characterization of asphalt concrete using image analysis. Journal of

Computing in Civil Engineering, ASCE 13(2), 88- 95 (1999 (a)) Special Issue on Image Processing.

13. Tashman, L., Masad, E., Peterson, B., Hebeeb, S.: Internal Structure

Analysis of Asphalt Mixes to Improve the Simulation of Superpave

Gyratory Compaction to Field Condition. Association of Asphalt Paving

Technologists 70, 605-645 (2002)

14. Wang, L. B., Paul, H. S., Harman, T., Anglo, J. D.: Characterization of aggregate and asphalt concrete using X ray tomography. (2004)

15. Landis, E., Keane, D.: X-ray micro-tomography for fracture studies in cement-based materials. International Society of Optical Engineering

3772, 105- 113 (1999)

16. Krishna, M., Rajagopol, K.: A brief review of the uses and modeling of bitumen from ancient to modern times. Internal Report, Department of

Mechanical Engineering, Texas A&M University, College Station. (2002)

17. Wang, L. B., Frost, J. D., Shashidhar, N.: Microstructure study of Westrack mixes from X-Ray tomography images. TRB Transportation Research

Record 1767, 85- 94 (2001 (a))

59

18. Oda, M.: Initial fabric and their relations to mechanical properties of granular material. Soils and Foundations 12(1), 17- 36 (1972)

19. Oda, M., Nemat-Nasser, S., Konish, J.: Stress-induced anisotropy in granular masses. Soils and Foundations 25(3), 85- 97 (1985)

20. Muhunthan, B., Chameau, J., Masad, E.: Fabric effects on the yield behavior of soils. Journal of Soils and Foundations 36(3), 85- 97 (1996)

21. Underwood, E. E.: Quantitative Stereology., Addison-Wesley Publishing

Company (1970)

22. Hilliard, J. E.: Specifications and measurement of microstructural anisotropy. AIME 224 (1962)

23. Hilliard, J. E.: Determination of structure anisotropy. In : Second

International Congress for Stereology, NY. (1967)

24. Kanatani, K.: Stereological determination of structural anisotropy.

International Journal of Engineering Science 22(5), 531- 546 (1984)

25. Kanatani, K.: Procedures for stereological estimation of structural anisotropy. International Journal of Engineering Science 23(5), 587- 598

(1985)

26. Synolakis, C. E., Zhou, Z., Leahy, R. M.: Determination of internal deformation field in asphalt cores using X-Ray computed tomography.

Transportation Research Record 1526, 135- 141 (1996)

27. Masad, E., Muhunthan, B., Shashidhar, N., Harman, T.: Internal structure characterization of asphalt concrete using Image analysis. (1999)

28. Wang, L., Frost, J. D., Shashidhar, N.: Microstructure study of Wes Track mixes from X-ray tomography images. (2001)

29. Tashman, L., Masad, E., Peterson, B., Habeeb, S.: Internal structure

60 analysis of mixes to improve the simulation of superpave gyratory compaction to field conditions. (2002)

30. Jönsson, M., Partl, M., Flisch, A.: Comparison of Laboratory Compaction

Methods using X-ray Computed Tomography. No. FE 840544 ISBN 3-

905594-31-5 (2007)

31. Masad, E., Jandhyala, V. K., Dasgupta, J., Samadeva, N., Shashidhar, N.:

Characterization of air void distribution in asphalt mixes using X-Ray CT.

Journal of Materials in Civil Engineering, ASCE 14(2), 122- 129 (2002 (a))

32. Wang, L. B., Frost, J. D., Voyiadjis, G. Z., Harman, T. P.: Quantification of damage parameters using X-ray tomography images. Mechanics of

Materials 35, 777- 790 (2003a)

33. Wang, L. B., Paul, H. S., Harman, T., D'Angelo, J.: Characterization of aggregates and asphalt concrete using X-ray tomography. Association of

Asphalt Paving Technologists 73, 467- 500 (2004b)

34. Tashman, L., Massad, E., Dallas, L., Hussein, Z.: A microstructure-based viscoplastic model for asphalt concrete. (2005)

35. Desruses, J., Chambon, R., Mokni, M., Mazerolle, F.: Void ratio evolution inside shear bands in triaxial sand specimen studied by computed tomography. Geotechnique 43(3), 529- 546 (1996)

36. Geraud, Y., Mazerolle, F., Raynaud, S., Lebon, P.: Crack location in granitic samples submitted to heating, low confining pressure and axial loading.

Geophysical International Journal 133, 553- 567 (1998)

37. Hall, C. C., Jupe, S. L., Jacques, A. C., Livingston, S., Ramadan, R., Made, A.,

Barnes, A., Paul, W.: Non-destructive tomographic energy-dispersive diffraction imaging of the interior of bulk concrete. Cement and Concrete

Research 30(3), 491- 495 (2000)

38. Yue, Z. Q., Bekking, W., Morin, I.: Application of Digital Image Processing

61 to Quantitative Study of Asphalt Concrete Microstructure. TRB

Transportation Research Record 1492, 53-60 (1995)

39. Hunter, A. E., Airey, G. D., Collop, A. C.: Aggregate Orientation and segregation in laboratory compacted asphalt samples. (2004)

40. Tashman, L., Masad, E., D'Angelo, J., Bukowski, J.: X-Ray tomography to characterize air void distribution in Superpave gyratory compacted specimens. International Journal of Pavement Engineering 3(1), 19-28

(2002 (a))

41. Liang, R. L., Shaw, H. L.: Anisotropic hardening plasticity model for sands.

Journal of the Geotechnical Engineering Division, ASCE 117(6), 913- 933

(1991)

42. Birgisson, B., Roque, R., Tia, M., Masad, E.: Development and Evaluation of Test Methods to Evaluate Water Damage and Effectiveness of

Antistripping Agents. Final Report 4910-4504-722-12, University of

Florida, Gainesville, FL (2005)

43. Masad, E., Niranjanan, S., Hussain, U., Sadi, K.: Modeling and experimental measurement of strain distribution in asphalt mixes. (2002)

44. Chen, W. F., Han, D. J.: Plasticity for structural engineers. Springer-Verlag

(1988) New York.

45. Nemat-Nasser, S., Mehrabadi, M.: Stress and fabric in granular mass.

Mechanics of Granular Materials: New Models and Constitutive Relations,

1- 8 (1983)

46. Tobita, Y.: Fabric tensors in constitutive equations for granular materials.

Soils and Foundations 29(4), 99- 104 (1989)

47. Oda, M., Nakayama, H.: Yield function for soil with anisotropic fabric.

Journal of Engineering Mechanics, ASCE 15(1), 89- 104 (1989)

62

48. Oda, M.: Inherent and induced anisotropy in plasticity theory of granular soils. Mechanics of Materials 16, 35- 45 (1993)

49. Muhunthan, B., Chameau, J. L.: Void fabric tensor and ultimate state surface of soils. Journal of the Geotechnical Engineering Division, ASCE

123(3), 173- 181 (1997)

50. Li, X. S., Defalias, Y.: Constitutive modeling of inherently anisotropic sand behavior. Journal of Geotechnical and Geoenvironmental Engineering

28(10), 868-880 (2002)

51. Konishi, J., Naruse, F.: A Note on Fabric in Terms of Voids.

Micromechanics of Granular Materials, 39- 46 (1987)

52. Kuo, C. Y., Frost, J. D.: Quantifying the fabric of granular materials-An image analysis approach. GIT-CEE/GEO-95-1, Department of Civil and

Environmental Engineering, The Georgia Institute of Technology, Atlanta,

GA (1995)

53. Watson, D., Masad, E., Moore, K., Williams, K., Cooley, A. J.: Verification of voids in coarse aggregate testing-Determining stone-on-stone contact of hot-mix asphalt mixtures. TRB, 182- 190 (2004) Transportation

Research Record 1891.

54. Curray, J. R.: Analysis of Two Dimensional Orientation Data. Journal of

Geology 64, 117- 131 (1956)

55. Al-Omari, A., Tashman, L., Masad, E., Cooley, A., Harman, T.: Proposed

Methodology for Predicting HMA Permeability. Journal of the Association of Asphalt Paving Technologists 71, 30-58 (2002)

56. Walsh, J. B., Brace, W. F.: The Effect of Pressure on Porosity and the

Transport Properties of Rocks., 9425- 9431 (1984)

57. Masad, E., Birgisson, B., Al-Omari, A., Cooley, A.: Analysis of Permeability and Fluid Flow in Asphalt Mixes. Journal of Materials in Civil Engineering,

63

American Society of Civil Engineers 16(5), 487- 496 (2004)

58. Tashman, L., Masad, E., Crowe, C., Muhunthan, B.: Simulation of fluid flow in granular microstructure using a non-staggered grid scheme.

International Journal of Computers and Fluids 32, 1299- 1323 (2003(a))

59. Casteloblanco, A., Masad, E., Birgisson, B.: Moisture Damage as a

Function of Air Void Size Distribution and Surface Energy. Journal of

Testing and Evaluation, American Society of Testing and Materials 34(1)

(December 2005) Online Paper ID: JTE13112.

60. Wang, L. B.: Stress concentration factor of poroelastic material by FEM simulation and X-ray tomography imaging. In : ASCE Engineering

Mechanics Conference, Seattle, WA (2003 (a))

61. Wang, Y. P., Wang, L. B., Abadie, C.: Non-invasive measurement of 3D permanent strains in asphalt concrete using X-ray tomography imaging.

Journal of Civil Engineering Materials, ASCE (2005b)

62. Mroz, Z., Norris, V. A., Zienkiewicz, O. C.: Applications of an anisotropic hardening model in the analysis of elasto-plastic deformation of soils.

Journal of Geotechnique 29(1), 1- 34 (1979)

63. Baker, R., Desai, C. S.: Induced anisotropy during plastic straining.

International Journal of Numerical and Analytical Methods in

Geomechanics 8(2), 167-185 (1984)

64. Masad, E., Muhunthan, B., Chameau, J. L.: Stress-strain model for clays with anisotropic void ratio distribution. International Journal for

Numerical and Analytical Methods in Geomechanics 22, 393- 416 (1998)

65. Birgisson, B., Fakimi, A., Wang, J.: Practical Modeling of Rock Mass

Behavior Under Dynamic Loading: Softening and Damage in Continuum

Models. Final Report DTRA 0003, Defense Threat Reduction Agency

(2006) Subcontract S03-36.

64

66. Satake, M.: Fundamental quantities in the graph approach to granular materials. Mechanics of Granular Materials: New Models and Constitutive

Relations, 9-19 (1983)

67. Boehler, J.-P.: Applications of tensor functions in solid mechanics. Volume

292 of Courses and lectures - International Centre for Mechanical

Sciences. University of Michigan (1987)

68. Kachanov, L. M.: On creep fracture time. Izv. Akad. Nauk USSR Otd. Tekh.

8 8, 26- 31 (1958) In Russian.

69. Murakami, S.: Notation of continuum damage mechanics and its application to anisotropic creep damage theory. Journal of Engineering

Materials and Technology 105, 99-105 (1983)

70. Lemaitre, J.: A continuous damage mechanics model for ductile fracture.

Journal of Engineering Materials and Technology 107, 83-89 (1985)

71. Zbib, H. M., Aifantis, E. C.: Instabilities during tension of thin voided viscoplastic sheets. Metallurgical Transactions 17(A), 1637- 1640 (1986)

72. Chaboche, J. L.: Continuum damage mechanics: Part I-general concepts.

Journal of Applied Mechanics 55, 59- 64 (1988)

73. Kattan, P. I., Voyiadjis, G. Z.: A coupled theory of damage mechanics and finite strain elasto-plasticity-I. Damage and elastic deformations.

International Journal of Engineering Science 28(5), 421- 435 (1990)

74. Voyiadjis, G. Z., Kattan, P. I.: A coupled theory of damage mechanics and finite strain elasto-plasticity-II. Damage and finite strain plasticity.

International Journal of Engineering Science 28(6), 505- 524 (1990)

75. Voyiadjis, G. Z., Kattan, P. I.: A plasticity-damage theory for large deformation of solids-I. The theoretical formulation. International Journal of Engineering Science 30(9), 1089- 1108 (1992)

65

76. Wang, L. B., Frost, J. D., Voyaiadjis, G. Z., Harman, T. P.: Quantification of damage parameters using X-ray tomography images. (2003)

77. 414-01, F.: Asfaltbeläggning och-massa. Bituminous pavement and mixture. Preparation of Marshall specimens. (2001)

78. ASTM-D-1557: Test Methods for Laboratory Compaction Characteristics of Soil Using Modified Effort. (1975)

79. Chapra, S., Canale, R.: Numerical Method for Engineers. (1985)

80. Kaloush, Witczak: Simple performance Test for Permanent Deformation of Asphalt Mixture. Annual Transportation Research Board Meeting,

Washington D.C. (2002)

81. Kamal, M. A., Shazeb, F., Baber, Y.: Resilient Behaviour of Asphalt

Concrete Under Repeated Loading and Effects of Temperature. Eastern

Asia Society for Transportation Studies, 1329-1343 (2007)

82. Tigdemir, M.: Dynamic Permanent Deformation Testing of Asphalt Mixes and Deformation Waveform Analysis. Indian jurnal of Engineering and

Materials Sciences, 29-35 (2008)

83. Kaloush, K., Withzak, M., Sullivan, B.: Simple Performance Test for

Permanent Deformation Evaluation of Asphalt Mix. In : 6th RILEM

Symposium PTEBM, Zurich, pp.489-505 (2003)

84. Fujie, Z., Scullion, T., Lijun, S.: Verification and Modeling of Three-Stage

Permanent Deformation Behaviour of Asphalt Mixes. Transportation

Engineering, 486-494 (2004)

85. AASHTO: Design of New and Reconstructed Flexible Pavements., Chapter

3 (2002)

86. Panoskaltsis, V.: Mechanics of Asphalt Concrete: Analytical and

Computational Studies., Cleveland, Ohio (2005)

66

87. Krishna, P. B., Kamil, E. K., Michael, S. M., Witczk, M. W.: Rational

Modeling of Tertiary Flow for Asphalt Mixture. Transportation Research

Board, 63-72 (2007)

88. M. W., W., B., S.: Superpave Support and Performance Models

Management. (2002)

89. VVTBT: Bitumenbundna lager 09 rev 2., Borlänge (2010)

90. Hubbell, J. H., Selzer, S. M.: Tables of X-Ray Mass Attenuation Coefficients and Mass Energy-Absorption Coefficients from 1 keV to 20 MeV for

Elements Z = 1 to 92 and 48 Additional Substances of Dosimetric Interest.,

National Institute of Standards and Technology. NISTIR 5632. (1996)

91. Ketcham, R. A., Carlson, W. D.: Acquisition, optimization and interpretation of X-ray computed tomographic imagery: Application to the geosiences. (2001)

92. Bonaquist, F. R., Christensen, D. W., Stump, W.: Simple Performance

Tester for Superpave Mix Design: First Article Development and

Evaluation. (2003)

93. Bonaquist, R.: Mastersolver Version 2.2. (2009)

94. Vieman, L., Hakim, H., Said, S.: Determination resilient modulus by indirect tensile test- precision trials- Styvhetsmodul Jämförande provning.

(2007)

95. Nilsson, R.: Applicering av en viskoelastoplastik materialmodell på svenska beläggningstyper., Skanska (2005)

96. Oscarsson, E.: Prediction of permanent deformations in asphalt concrete using the Mechanistic-Empirical Pavement Design Guide., Lund University

(2007)

97. NCHRP-Report_547: Simple Performance Tests: Summary of

67

Recommended Methods and Database., Washington D.C. (2005)

98. Ketcham, R. A., Carlson, W. D.: Achuisition, optimization and interpretation of X-ray computed tomographic imagery: Applications to the geoscience. Computers and Geosciences 27, 381- 400 (2001)

99. Masad, E. A., Muhunthan, B., Shashidhar, N., Harman, T.: Quantifying

Laboratory Compaction Effects on the Internal Structure of Asphalt

Concrete., Washington D.C. (1999)

100. Raaberg, J.: Investigation of Gyratory Compaction used for Asphalt Mix

Design., Roskilde (1999)

101. Thyagarajana, S.: The heterogeneity and mechanical response of hot mix asphalt laboratory specimens. (2010)

102. Nillson, R.: Fatigue of Asphalt Mixture- Continuum Damage Mechanics applied to Data from Laboratory Tests., Lund (2001)

103. Chehab, G., Galal, K. A.: The Mechanistic-Empirical Design Guide: A case study on HMA overlay over Fracture PCC slab. (2005)

104. Tashman, L., Masad, E., Little, D., Lytton, R.: Damage evolution in triaxial compression tests of HMA at high temperatures. Association of Asphalt

Paving Technologists 73, 53- 87 (2004)

105. Menq, F.: Dynamic properties of Sandy and Gravelly Soils., Austin (2003)

106. Gichul Kweon, R.: Determination of Asphalt Concrete Complex Modulus

With Impact Resonance Test. Transport Research Board, 151-160 (2006)

107. Gudmarsson, A., Ryden, N., Birgisson, B.: Application of resonant acoustic spectroscopy to asphalt concrete beams for determination of the dynamic modulus., Turkey (2011)

108. NCHRP-Report_580: Specification Criteria for Simple Performance Tests

68 for Rutting. Volume I: Dynamic Modulus, Volume II: Flow Number and

Flow Time., Washington D.C. (2007)

109. Surrey, M., S.: The Shell Bitumen Industrial Handbook., UK. (1995)

110. Whitmoyer, S. L., Kim, Y.: Determining Asphalt Concrete Properties via the

Impact Resonant Method. (1994)

111. Saadeh, S. L., Masad, E., Mogawer, W.: Spatial and Directional

Distributions of Aggregates in Asphalt Mixes. (2002)

112. Tashman, L., Wang, L. B., Thyagarajan, S.: Microstructure characterization for modeling HMA behavior using imaging technology. (2007)

113. Song, I.: Damage analysis in asphalt concrete mixtures based on parameter relationships. (2004)

114. Masad, E., Hussain, B., Sadi, K.: Modeling and experimental measurements of Strain distribution in asphalt mixes. (2001)

115. Papagiannakis, A., Abbas, A., Masad, E.: Modeling the viscoelastic bevavior of asphalt concretes. (2002)

116. Kose, S., Guler, M., Bahia, H., Masad, E.: Distribution of strains within hot/mix asphalt binder. (2000)

117. Tashman, L.: Microstructural viscoplastic continuum model for asphalt concrete. (2003)

118. Dessouky, E., Masad, E., Bayomy, F.: Evaluation of asphalt mix stability using compaction properties and aggregate structure analysis. (2003)

119. Kutay, M. E., Arambula, E., Gibson, N., Youtcheff, J.: Three-Dimensional

Image Processing Methods to Identify and Characterize Aggregates in

Compacted Asphalt Mixtures. (2010)

69

120. Kim, Y. R., Lee, H., Little, D.: Fatigue characterization of asphalt concreta using viscoelasticity and continuum damage theory. (1997)

121. Nelson, H. G., Charles, G., Richard, S.: Viscoelastic, Viscoplastic and

Damage Modeling of Asphalt Concrete in Unconfined Compression.

(2003)

122. Bonaquist, R. F., Christensen, D. W., Stump, W.: NCHRP Report 513-

Simple Performance Tester for Superpave Mix Design: First- Article

Development and Evaluation. (2003)

123. Bonaquist, R., Christensen, D.: A practical procedure for developing dynamic modulus master curves for permanent structural design. (2005)

124. NCHRP-Project_1-37A, Associates, A.: Guide for Mechanistic-Empirical

Design of New and Rehabilitated Pavement Structures., Washington DC

(2004)

125. Wang, L.: Mechanistics of Asphalt: Microstructure and Micromechanics.

McGraw-Hill Professional (2010)

126. Cheng, D.: Surface Free Energy of Asphalt-Aggregate System and

Performance Analysis of Asphalt Concrete. Ph.D. Thesis, Texas A&M

University, College Station, Texas (2002)

127. Hunter, A. E., Airey, G. D., Collop, A. C.: Aggregate Orientation and

Segregation in Laboratory–Compacted Asphalt Samples., Transportation

Research Record 1891, Washington, D.C. (2005)

128. Witczak, K.: Simple performance Test for Permanent Deformation of

Asphalt Mixture. Annual Transportation Research Board Meeting,

Washington D.C. (2002)

129. Bossi, R. H., Knutson, B. W.: The advanced development of X-ray computed tomography applications. United States Air Force Wright

Laboratory Publication (1994)

70

130. ASTM-C-215: Standard test method for fundamental transverse longitudinal and torsional resonant frequency of concrete specimens.

131. Francken, L.: Bituminous Binders and Mixtures., London (1998)

132. Witczak, M. W., Sulivan, B.: Superpave Support and Performance Models

Management. (2002)

133. Ryden, N.: Determining the asphalt mastercurve from free-free resonant testing on cylindrical samples. In : NDTCE, Non-Destructive Testing in Civil

Engineering, Nates, France (2009)

134. Roberts, F. L., Kandhal, P. S., Brown, E. R., Lee, D. Y., Kennedy, T. W.: Hot

Mix Asphalt Materials, MixtureDesign and Construction., Lanham (1996)

135. Witczak, M. W.: Development of a mastercurve database for lime modified asphaltic mixtures. Arizona State University, Department of Civil and Environmental Engineering (2004)

136. FAS_Method: Asfaltbeläggning och-massa. Bituminous pavement and mixture. Preparation of Marshall specimens. (2001)

137. Lytton, R. L., Chen, C. W., Little, D. N.: Microdamage Healing in Asphalt and Asphalt Concrete, Volume III: A Micromechanics Fracture and Healing

Model for Asphalt Concrete. FHWA-RD-98-143, Texas Transportation

Institution, College Station, TX (2001)

138. Masad, E., Tashman, L., Niranjanan, S., Little, D.: Micromechanics based analysis of stiffness anisotropy in asphalt mixtures. Journal of Materials in

Civil Engineering, ASCE 14(5), 374- 383 (2002 (b))

139. Schapery, R. A.: Correspondence Principles and a Generalized J Integral for Large Deformation and Fracture Analysis of Viscoelastic Media.

International Journal of Fracture 25, 195- 223 (1984)

140. Lytton, R., Tashman, L., Masad, E., Little, D., Lytton, R.: Damage Evolution

71 in Triaxial Compression Test of HMA at High Temperatures. Association of

Asphalt Paving Technologists 73, 53-87 (2004)

141. Wang, L. B.: 3D FEM simulation for WesTrack mixes from X-ray tomography images. Research Report, School of Civil and Environmental

Engineering, Georgia Institute of Technology (1999)

142. Wang, L. B., Frost, J. D., Lay, J. S.: Three dimensional digital representation of granular material microstructure from X-ray tomography imaging.

Journal of Computing in Civil Engineering 18(1), 28- 35 (2004(a))

143. Wang, L. B., Wang, X., Mohammad, L. M., Wang, Y. P.: Evaluation of the mechanical properties of asphalt concrete using mixture theory. Journal of Materials in Civil Engineering, ASCE 16(2), 167- 174 (2003b)

144. MathWorks: http://www.mathworks.com.

Appendix

The appendix contains important information that complements the thesis.

A Simple Performance Tester (SPT)

A.1 SPT- test procedures

General procedure steps for SPT

To ensure the tests and the results are ran properly, the following checklist should be followed:

1. Run a dummy samples to ensure the machine is properly calibrated and installed.

2. The compressed air pressure supplied to the machine has to be in the required ranges; Use a air pressure gauge to verify.

3. Check the sample dimensions using high accuracy measuring instruments to verify the angles and to inspect wall surface weaviness. These verifications are very important as imperfect samples will highly affect the tests results.

4. Careful choose the glue (epoxy) used works yours materials, it hardens quickly (approximately 5 minutes) and resists at the temperatures to be used during the testing.

5. The horizontal faces of the studs have to be parallel when glued (so that the supports for holding the LVDTs are parallel and the LVDT is perfectly vertical).

6. When fixing the LVDT sensors check that the cables do not touch the chamber or the sample because otherwise errors can be introduced.

7. The LVDTs have to be adjusted properly for each test (close to zero or close to maximum range). This can be achieved via the screws in the sensors. One has to be careful as they are rather sensitive.

74

Dynamic Modulus test procedures

For Dynamic modulus tests these steps should be followed:

1. Ensure that the sample is perfectly centered on the loading plates on both sides.

2. The metallic ball has to be centered and aligned with the upper plate.

3. Place and adjust the LVDTs using the ranges shown in the Dynamic

Modulus software. Make sure all the LVDT sensors are as close as possible to 0. If you run into issues you can rotate the sample and try to switch between places in which you mount the sensors. When you are finished you should set the sensor readings from the software GUI to zero.

4. Make sure you properly condition the sample to the test temperature (for temperatures > 10°C, approximately 2 hours; for temperatures bellow 0°C minimum 8 hours in the environmental chamber).

5. The sample diameter and height must be precisely measured in several locations (6 diameter measurements and 3 for height) with a precision of

0.01 millimeters.

6. To ensure proper function of the equipment the sample has to be conditioned in the SPT environmental chamber for at least 20 minutes before starting the tests. In this time the operator can setup the test.

7. In the software give proper names and comments to the tests you are running, including the sample code or number, the date and time, the test specifics, etc.

8. Choose the frequencies you require from the machine in the GUI.

9. Run the test.

10. Inspect the results and verify that they are meeting the specifications

75

Flow Number (repeated load test) procedures

For this test, the following steps should be followed:

1. The tests have to be run at a temperature range between 30°C and 40°C. It is recommended that the samples are placed in the external environmental chamber for at least 4 hours.

2. Experience might help to determine the needed deviatoric stress and the contact stress. In this project trial tests were performed to establish the appropriate stresses (30kN for deviatoric stress and 70kN for contact stress).

3. End friction reducers should be placed above and below the specimen.

Usually, latex membranes sheets separated by silicon grease are used.

4. Because the samples are heated they start to soften and one has to make sure that they use the proper amount of glue so that the studs do not move. Try to rotate the sample so that the studs are glued to aggregates in the mix and as little as possible to the mastic

5. The LVDTs must be setup at maximum negative range (-0.5mm)

6. Setup parameters in the software interface (contact and deviator stress and termination settings: maximum microstrain or test duration)

7. Run the test and make sure you observe it because if the sample breaks the

LVDTs fall and can be damaged. In case of emergency stop the test.

76

77

A.2 SPT- Settings

Table 7. Flow Number Assumptions and SPT system features.

Assumption

Representative volume element

Consistent pulse loading

Consistent rest period

Uniform Stress state

SPT System Feature

100 mm diameter h/d ratio of 1.5

Load standard error

Tolerance on maximum load

Computer control

Specimen size, h/d=1.5

Smooth parallel ends and loading platens

Friction reducer

Table 8. Data quality statistics requirements for Dynamic Modulus measured with SPT.

Data Quality Statistics

Deformation Drift

Peak to Peak Strain

Load standard error

Deformation standard error

Deformation uniformity

Phase uniformity

Limit

In direction of applied load

75 to 125

µstrain unconfined tests

85 to 115

µstrain confined tests

10 %

10 %

30 %

3 degrees

Table 9. Statistical analysis of influence in sample rotation.

ANOVA

Variationsursprung

Mellan grupper

Inom grupper

KvS

44563.01

fg MKv

21360323.55 12 1780027

F p-värde F-krit

2 22281.51 0.01 0.98 3.88

Totalt 21404886.57 14

Because F is much smaller then F-crit value, the differences between measurements is not statistically significant.

78

79

A.3 Dynamic Modulus Master Curve equations

In the following, the workbook equations developed by NCHRP is presented and the parameters and master curve are obtained. The workbook is used in conjunction with the Simple Performance Test System to develop dynamic modulus master curves. It has the capability to solve a modified version of the

Mechanistic-Empirical Design Guide master curve equation, Equation 1. log

E

*

= log(

Min

)

+

( log(

Max

1

+

)

e

β

+

γ log( log

ω

r

Min

)

)

(1) where:

E* = dynamic modulus

ω

r

= reduced frequency, Hz

Max = limiting maximum modulus, ksi

Min = limiting minimum modulus, ksi

β

, and

γ

= fitting parameters

The reduce frequency is computed using the Arrhenius equation, Equation 2. log

ω

r

= log

ω

+

E a

19 .

14714



1

T

1

T r



(2) where:

ω

r

= reduced frequency at the reference temperature

ω

= loading frequency at the test temperature

T r

= reference temperature,

°K

T = test temperature,

°K

E

a

= activation energy (treated as a fitting parameter)

Substituting Equation 2 into Equation 1 yields the form of the master curve equation that is fitted using this workbook.

log

E

*

= log(min)

+

(

log(

1

+

e

β

+

γ

Max

)

− log(

Min

)

) log

ω

+

E a

19 .

14714

1

T



1

T r



(3)

80

The shift factors for each temperature are given by Equation 4. log

[

a

(

T

)

]

=

E a

19 .

14714



1

T

T

1

r



(4) where:

a(T) = shift factor at temperature T

T r

= reference temperature,

T = test temperature,

°K

°K

E

a

= activation energy (treated as a fitting parameter)

The maximum limiting modulus is estimated from mixture volumetric properties using the Hirsch model and a limiting binder modulus of 1 GPa

(145,000 psi), Equations 5 and 6.

|

E

* | max

=

P c

4 , 200 , 000

 −

VMA

100

+

435 , 000

VFA

10

x VMA

, 000

+

VMA

100

4 , 200 , 000

1

+

P c

VMA

435 , 000 (

VFA

)

(5) where:

0 .

58

20

+

435 , 000 (

VFA

)

VMA

P c

=

0 .

58

650

+

435 , 000 (

VFA

)

VMA

E* max

= limiting maximum mixture dynamic modulus

VMA = Voids in mineral aggregates, %

VFA = Voids filled with asphalt, %

The resulted master curve is presented in Figure 39.

(6)

81

Figure 39. Dynamic Modulus [ksi] vs. Reduced Frequency [deg] shifted - at different temperatures.

82

83

Table 10. The SPT results and Master Curve calculation.

Temp.

10

10

10

1

1

1

1

1

-5

-5

-5

1

1

1

1

°C

-5

-5

-5

-5

-5

-5

20

20

20

20

20

20

20

35

35

35

35

35

35

35

35

35

10

10

10

10

10

10

20

20

Project:

Mix

Date:

Technician:

Samples

VMA

VFA

Ref. Temp

Conditions

Freq.

2

5

10

20

25

0.1

0.2

0.5

10

20

25

0.1

0.2

0.5

1

Hz

0.1

0.2

0.5

1

2

5

0.5

1

2

5

10

20

25

0,1

0.2

0.5

1

2

5

10

20

25

1

2

5

10

20

25

0.1

0.2

Volume, %

Volume, %

20

Sample DM1

Modulus

Phase

Angle

Licentiate Thesis

AG22, 70/100

2011-10-22

Florentina Farcas

Specimen 1

13,48

58,11

°C

Sample DM2

Modulus

Phase

Angle

Specimen 2

14,14

60,39

Average

Modulus

Ksi

1568.9

1759.3

2010.5

2200.1

2379.4

2605.2

2761.7

2919.2

2949.8

1068.1

1263.2

1531.2

1740.2

1959.1

2247.8

2466.7

2679.0

2732.8

393.3

501.8

673.9

811.4

963.4

1190.1

1371.6

1569.3

1633.4

72.9

108.1

175.4

245.4

340.1

487.8

621.8

764.9

801.8

10.1

13.1

19.9

28.8

42.8

81.4

125.3

184.2

207.7

Degree

16.04

14.50

12.56

11.27

10.13

8.79

7.95

7.25

6.77

22.08

20.33

18.03

16.45

14.96

13.05

11.65

10.35

9.68

33.62

31.47

28.39

38.64

37.41

35.65

32.94

30.78

28.57

28.68

27.83

31.80

36.40

39.23

26.14

24.04

21.29

19.40

17.57

17.12

38.90

39.00

41.73

40.87

40.55

40.37

40.34

Ksi

1794.0

1984.9

2242.2

2452.5

2651.0

2920.2

3126.3

3288.9

3355.1

1144.9

1303.3

1529.0

1712.5

1902.2

2158.8

2351.2

2544.9

2596.5

463.5

588.1

773.9

934.1

1100.4

1329.1

1521.5

1725.7

1791.4

153.3

197.1

284.0

366.4

462.5

625.1

772.3

930.7

985.7

13.4

18.0

27.5

38.8

61.1

111.9

169.8

248.0

276.7

Degree

15.53

14.00

12.09

10.72

9.69

8.36

7.56

7.08

7.17

22.09

20.37

18.01

16.35

14.78

12.90

11.68

10.72

10.10

31.72

29.63

26.76

34.69

34.06

32.60

30.46

28.59

26.66

26.63

28.51

31.49

35.35

38.50

24.77

22.80

20.33

18.62

16.88

16.72

32.69

34.12

39.24

38.68

38.11

37.65

38.31

Ksi

1681.4

1872.1

2126.4

2326.3

2515.2

2762.7

2944.0

3104.1

3152.4

1106.5

1283.2

1530.1

1726.3

1930.6

2203.3

2409.0

2612.0

2664.7

428.4

545.0

723.9

872.7

1031.9

1259.6

1446.6

1647.5

1712.4

113.1

152.6

229.7

305.9

401.3

556.4

697.1

847.8

893.7

11.8

15.6

23.7

33.8

51.9

96.7

147.5

216.1

242.2

Average

13,8

59,3

Average DM1 DM2

Phase Modulus

Degree MPa

15.8 10817

14.3 12130

12.3 13862

11.0 15169

9.9 16405

8.6 17962

7.8 19041

7.2 20127

7.0 20338

22.1 7364

20.4 8709

18.0 10557

16.4 11998

14.9 13507

13.0 15498

11.7 17007

10.5 18471

9.9 18842

32.7

30.6

27.6

2712

3460

4646

36.7

35.7

34.1

31.7

29.7

27.6

27.7

28.2

31.6

35.9

38.9

25.5

23.4

20.8

19.0

5594

6642

8205

9457

17.2 10820

16.9 11262

35.8

36.6

503

746

40.5

39.8

39.3

39.0

39.3

295

562

864

1270

1432

1209

1692

2345

3363

4287

5274

5528

70

90

137

199

MPa

12369

13685

15459

16909

18278

20134

21555

22676

23132

7894

8986

10542

11807

13115

14884

16211

17546

17902

3196

4055

5336

6440

7587

9164

10490

11898

12351

1057

1359

1958

2526

3189

4310

5325

6417

6796

93

124

190

268

421

771

1171

1710

1908

84

B X-ray CT Results

85

Figure 40. X-ray CT image - sample CT3.

86

87

Figure 41. Aggregates in the 3D volume of interest- sample CT1.

88

89

90

Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement

Table of contents