Read Characterization of Hot Mix Asphalt with Varying Air Voids Content Using Triaxial Shear Strength Test text version

CHARACTERIZATION OF HOT MIX ASPHALT WITH VARYING AIR VOIDS CONTENT USING TRIAXIAL SHEAR STRENGTH TEST

T.K. Pellinen, J. Song and S. Xiao

School of Civil Engineering 550 Stadium Mall Drive, West Lafayette, IN 47907-2051. E-mail: [email protected], [email protected] and [email protected]

ABSTRACT

Two asphalt mixtures, a dense graded mix (DGM) and a Stone Mastics Asphalt (SMA) were investigated using triaxial shear strength testing. The objective was to study the effect of air voids content on the mixture strength properties by varying air voids content from 0 to 13%. The Mohr-Coulomb failure theory was used to obtain cohesion and friction angle. Analysis of test data shows that both mixtures have highest cohesion, unconfined compressive strength and shear strength at zero percent air voids content. Testing was conducted at 50 mm/min ram rate. Furthermore, the study findings suggest that the strength of the SMA mixture is less sensitive for the variation of air voids content compared to the DGM. Application of Critical State Soil Mechanics theory suggests that the SMA mixture has less potential for dilatation compared to the DGM when pseudo critical state at 10% axial strain was used in the analysis. Keywords: Hot Mix Asphalt (HMA), Triaxial Strength Test, Loading Rate, Air Voids Content, Permanent Deformation, Critical State Soil Mechanics

1. INTRODUCTION

Performance of asphalt mixtures is dependent on the volumetric composition of the mixture, in addition to the raw materials used. Volumetric composition includes volume of aggregate and binder, and volume of air. Research has shown that the stiffness of the mixture is highly dependent on the air voids content of the mixture, which is recognized in all stiffness predictive models such as the Shell model by Bonnaure et al., the Witczak et al. model by Andrei et al., and the Hirsch model by Christensen et al. [1, 2, 3, 4]. There are no similar models for the strength of the mixture, although triaxial shear testing has been used for material characterization steadily in recent years [5, 6, 7, 8, 9, 10]. However, research shows that both mechanical properties stiffness and compressive strength are increasing when the air voids content of the mixture is decreasing while the binder content is kept constant. In other words, the mixture is getting stronger and stiffer when the air voids content decreases. In the U. S. the as-constructed air voids content usually deviates the most from the original mixture properties determined by the laboratory Superpave Mix Design method. The objective of the research is to investigate how stability of asphalt mixtures is affected by the varying air voids content in the mixture. Stability can be defined as the mixture's ability to resist plastic flow when loaded. Stability of the mixture was obtained from the triaxial testing. The test data is analyzed using conventional Mohr-Coulomb theory but also a novel approach of using Critical State Soil Mechanics (CSSM) theory has been pursued. Due to the preliminary nature and limitations of the laboratory testing, some caution when applying the CSSM theory must be used while assessing test results. The scope is to estimate and compare behavior of dense graded asphalt mixture (DGM) and Stone Mastic Asphalt Mixture (SMA).

Proceedings of the 8th Conference on Asphalt Pavements for Southern Africa (CAPSA'04) ISBN Number: 1-920-01718-6 Proceedings produced by: Document Transformation Technologies cc

12 ­ 16 September 2004 Sun City, South Africa

8th CONFERENCE ON ASPHALT PAVEMENTS FOR SOUTHERN AFRICA

It should be noted that mixture performance such as permanent deformation is usually associated with the increase in voids filled with bitumen. However, it is important to separate the two different ways of filling the voids in asphalt mixture: increasing compaction while keeping the binder content constant or increasing binder content while keeping the compaction constant. The later method is used in the Marshall Mix Design method to verify the optimum binder content relative to the standard compaction of 35, 50 or 75 blows of Marshall Hammer. This paper investigates the first method which relates to the in-situ mix performance through the mechanical properties of asphalt mixture relative to the in-situ construction air voids content and the consequent in-service traffic loading.

2. LABORATORY EXPERIMENT 2.1 Mix Stability

Stability of asphalt mixtures was studied heavily in the 40's and 50's and stability mix design methods based on the triaxial shear strength testing were developed by many researchers [11, 12, 13, 14]. Although the Marshall Stability test is considered as an index test, it gained popularity over the other more fundamental methods [15] and it has been the most widely used mix design method in the world today. Therefore, due to the preliminary nature of this study, a single loading rate and test temperature close to the Marshall Stability testing has been selected for this research. However, because loading rate greatly affects mechanical properties of asphalt as research shows [16], a separate loading rate study will be conducted later as a continuation of this experiment. The mix stability in the form of shear strength obtained from the triaxial testing is considered as a fundamental material property that could be used in the mechanistic pavement design procedures. Triaxial shear strength testing has traditionally been conducted at 60°C temperature using very slow loading rates such as strain rate of 0.05 mm/mm/min to minimize the binder stiffness effect on mixture cohesion during testing [11, 12, 13, 14]. This strain rate is equivalent to a 7.5 mm/min ram rate for a 150 mm high specimen. The Marshall Stability test is conducted at 60°C using 50 mm/min ram rate according to the specification.

2.2 Tested Mixtures

Two different mixture types have been tested in this study, a SMA mixture and a dense graded mixture. These mixtures are typical interstate surface mixtures used by the Indiana Department of Transportation. The dense grade mixture was placed on State Route 135 and the SMA mixture on US 31 during summer 2002. Test specimens have been prepared using loose mixture obtained from the asphalt plant during construction. Figure 1 gives the gradation and Superpave mix design Job Mix Formula for the studied mixtures. The (Pb) is the total binder content by weight and (Vbeff) is the effective binder content by volume, excluding the absorbed binder at 4% design air voids content. The aggregate blend in the SMA was composed mostly of steel slag having specific gravity greater than 3. Therefore, by weight the SMA binder content was only 4.7% compared to the 6.0% in the DGM, while compared by volume basis the SMA had 13.8% and the DGM had 11.4% of binder. The SMA mixture did not have fibers but it had modified PG 76-22 binder, while the dense graded mixture had PG 70-22 binder. Percent passing 0.075 mm sieve for the DGM was 4.4% and for the SMA 8%. Dust/asphalt ratios were 0.9 and 1.7, respectively. Both mixtures were designed with Ndes = 100 gyrations.

Paper 027

8th CONFERENCE ON ASPHALT PAVEMENTS FOR SOUTHERN AFRICA

Figure 1. Tested mixtures and their mix design properties.

2.3 Specimen Fabrication and Test Conditions

The loose plant mix was reheated in the laboratory and then compacted with Superpave Gyratory Compactor (SGC). All samples were cored from 150-mm diameter by 172-mm high gyratory compacted pills and saw-cut to obtain parallel end faces. The final sample size was 100 mm in diameter and 150 mm in height based on recommendations by Witzcak et al. [17]. The study was conducted using four target air voids content of 0, 4, 8 and 12%. The reason for selecting 8 and 12% air voids content was to cover the medium and upper range of in-situ air voids content encountered in mixtures in the U. S. The zero air voids content was selected to cover the refusal density of the mix or the ultimate state of the mix, and 4% is the design air voids content used in the Superpave volumetric mix design system. Figure 1 shows that the gradation of DGM was quite close to the maximum density line (Fuller curve), which allowed dense packing of aggregate particles while applying maximum number of gyrations (400) by the SGC. Also, the SMA mixture gradation allowed densification to the zero air voids content. Table 1 shows the target air voids content (Va), actual measured air voids content, and consequent volumetric properties, voids in mineral aggregate (VMA), voids filled with asphalt (VFA), and effective binder volume (Vbeff) of the specimens for both mixtures. The VFA ranged from 44.1 to 100% and VMA from 11.9 to 24.8%. Table 1 also shows the number of gyrations needed to compact mixtures to the desired air voids content. The number of gyrations refers to the compaction of the 150 x 172 mm gyratory pills from which the actual test specimens were cored. The gyratory compact samples are denser from the middle which must be compensated during specimen fabrication by increasing the average target air voids content for the pills. The target in Table 1 refers to the air voids content of the cored test specimens. When compared relatively, the amount of compaction needed in the cored specimens to achieve the zero air voids content was 80 times more than to achieve 12% air voids, and 12 time and 4 times more than to achieve 8% and 4% air voids contents, respectively. Test specimens were sealed using a latex membrane but no instrumentation was attached to the specimens nor used during testing. Lubricated disks of membrane were used between the sample and platens to prevent friction during loading. Four different confinement pressures of 0, 138, 276, and 414 kPa were used in the testing. Testing was conducted by testing one specimen at each confining pressure. The test temperature and loading rate (ram rate) were 55°C and 50 mm/min, respectively. The test temperature was selected to match the test temperature for the dynamic modulus testing which was planned to be conducted later on. The specimens were loaded up to axial strain of 30% to obtain post peak behavior. Two alternative termination criteria were used to abort the test: either

Paper 027

8th CONFERENCE ON ASPHALT PAVEMENTS FOR SOUTHERN AFRICA

to retain 20% of the peak load or to reach 15 to 30% axial strain. Due to the exploratory nature of the research, test specimens were not instrumented, as mentioned above, and volumetric strains could not be measured during testing. However, volume change of selected specimens was measured before and after testing using Corelok vacuum system [18] to verify visual observations during testing. The testing apparatus included geotechnical cell with air confinement and 25 kN loading frame which were mounted to the temperature controlled environmental chamber.

Table 1. Volumetric composition of tested mixtures.

Mix Type

Target Va (%)

12 8 4 0 12 8 4 0

Measured VMA (%) Va (%)

13.2 8.3 3.9 0.0 12.1 7.9 3.4 0.0 23.6 19.2 15.4 11.9 24.8 21.2 17.3 14.4

VFA (%)

44.1 56.8 74.7 100.0 51.2 62.7 80.3 100.0

Vbeff (%)

10.4 10.9 11.5 11.9 12.7 13.3 13.9 14.4

Avg. No. of Gyrations

5 19 65 400 5 22 56 400

DGM

SMA

3. LABORATORY TEST RESULTS 3.1 Stress-Strain Curves for Dense Graded Mixture

Figure 2 shows the measured stress-strain curves for the dense-graded mixture for each air voids content and confinement level (3). Both variables have an effect on the shape and type of the stress-strain curves. The failure of the mixture was obtained as a peak (maximum) deviatoric stress (d = 1 - 3). The strength of the zero air voids content specimen tested at 414 kPa confinement was not measured because it exceeded the capacity of the 25 kN loading frame, see Figure 2(d). In addition, test results for the zero air voids content specimen with 138 kPa confinement was considered erroneous and was excluded, Figure 2(b). Testing was not repeated due to the lack of material. Overall, the stress-strain curves were fairly linear at the beginning of the test. Figure 3 shows a close up of the stress-strain curves for the mixture with zero and 13% air voids content tested at 276-kPa confinement. The shapes of the stress-strain curves indicate a different failure strain and behavior for these two specimens. The DGM has a typical stress-strain curve of a brittle type material with high stiffness, while the SMA mixture showed a typical plastic type failure with large "necking" with lower stiffness. A photo of the zero air voids content specimen at 4% post peak axial strain also suggests that the failure occurred as brittle type cracking and it is difficult to say if there is a clear shear failure plain in the specimen. At this point the residual stress declined to approximately 19% of the peak (failure) deviatoric stress of 2,602 kPa. A photo of the specimen with 13% air voids at post peak axial strain of 26% shows no clear shear failure plain and specimen was bulged from the middle. Visual observation during testing confirmed that bulging was initiated at post peak i.e., after the specimen had failed, and bulging gradually increased as the axial strain was continuously increased. Specimen bulging occurred when mixture tried to expand laterally but was restricted by the latex membrane fixed to the platens by o-rings. At 26% axial strain the residual stress was approximately 66% of the peak (failure) deviatoric stress of 1,504 kPa.

Paper 027

8th CONFERENCE ON ASPHALT PAVEMENTS FOR SOUTHERN AFRICA

Va = 0% Va = 8% 2000 d , kPa 1500 1000 500 0 0 5

Va = 4% Va = 13% 2000 d , kPa 3 = 0 kPa 1500 1000 500 0 0

Va = 4% Va = 13%

Va = 8%

3 = 138 kPa

(a)

10 15 20 25 30 35 Axial Strain (%) Va = 4% Va = 13% 3 = 276 kPa

(b)

5

10 15 20 25 30 35 Axial Strain (%) Va = 8%

Va = 0% Va = 8% 3000 2500 2000 1500 1000 500 0 0 5

Va = 4% Va = 13% 3000 2500 2000 1500 1000 500 0 0 5

d , kPa

(c)

10 15 20 25 30 35 Axial Strain (%)

d , kPa

3 = 414 kPa

(d)

10 15 20 25 30 35 Axial Strain (%)

Figure 2. Stress-strain curves for the DGM for various confinement levels.

Note that the sample sizes in the photos are not in scale. At post peak the diameter of the zero air voids content specimen increased approximately 4% while the average diameter increase in the other specimen was approximately 20% at the end of the test. At this point the specimen volume increase was approximately 4% and 10%, respectively, based on measurements done by Corelok vacuum system [18]. Because the zero air voids content specimen failed without bulging it suggests that the friction prevention method used was quite effective. As mentioned earlier, post peak bulging of the specimen with 13% air voids content was caused by large lateral movement partially restricted by the membrane around the specimen.

Va = 0%

Va = 13%

3000 2500 2000 1500 1000 500 0 0 5 10 15 20 25 Axial Strain (%) 30 3 = 276 kPa

Va = 0% a = 4% Dia. Increase 4%

d , kPa

Va = 13% a = 26% Dia. Increase 20%

Figure 3. Post peak stress-strain curves and specimen shape after testing for DGM.

Paper 027

8th CONFERENCE ON ASPHALT PAVEMENTS FOR SOUTHERN AFRICA

The stress-strain relationships in Figure 2 suggest that with the same confinement level mixes with low air voids content exhibit brittle type failure characteristics while the mixes with high air voids content exhibit plastic type failure characteristics. Furthermore, the higher confinement reduces the brittle type failure characteristics and increases the failure strain. These results are in agreement with the findings reported by Schnaid, Prietto, and Consoli [19] for cemented sand in which the cohesion intercept is rising from the chemical bonding of sand particles.

3.2 Stress-Strain Curves for SMA Mixture

Figure 4 shows measured stress-strain curves for the SMA mixture. The shape change of the stress-strain curves is more gradual and brittle type failure does not seem to occur as clearly as for the DGM. Also, the failure strains are slightly higher ranging from 3 to 21% compared to 2 to 8% for the DGM.

Va = 0% Va = 8%

Va = 4% Va = 12%

Va = 0% Va = 8%

Va = 4% Va = 12%

2000

d , kPa

d , kPa

1500 1000 500 0 0 5

3 = 0 kPa

2500 2000 1500 1000 500 0 0 5

3 = 138 kPa

(a)

10 15 20 25 30 35 Axial Strain (%)

Va = 4% Va = 12%

(b)

10 15 20 25 30 35 Axial Strain (%)

Va = 4% Va = 12%

Va = 0% Va = 8%

Va = 0% Va = 8%

d , kPa

1000 500 0 0 5 10 15 20 25 30 35 Axial Strain (%)

d , kPa

2500 2000 1500

3 = 276 kPa

3500 3000 2500 2000 1500 1000 500 0 0 5

3 = 414 kPa

(c)

(d)

10 15 20 25 30 35 Axial Strain (%)

Figure 4. Stress-strain curves for the SMA for various confinement levels.

Figure 5 shows a close up of the stress-strain curves for the SMA mixture with zero and 12% air voids content. After testing, the zero air voids content specimens seem to have less visible cracks and the 12% specimen seem to be less bulged compared to the DGM. The diameter increase was approximately 8% for the dense mixture and 20% for the loose mixture. The volume change was approximately 6% and 10%, respectively. The stress-strain curve in Figure 5 suggests that the 12% mixture had a large movement in the aggregate skeleton during testing. The following phenomena were observed during testing based on visual observations and inspection of stress-strain curves and failed specimens for both the DGM and SMA mixtures at unconfined stress state: For the DGM at around 0.7% axial strain (approximately 1 mm axial deformation) and for the SMA at around 0.5% axial strain aggregates started to slip initiating a structural transformation. This process continued until aggregate particle interlock was overcome and dilatation (volume increase) took place in the specimen.

Paper 027

8th CONFERENCE ON ASPHALT PAVEMENTS FOR SOUTHERN AFRICA

Va = 0%

Va = 12%

3000 2500

d , kPa

3 = 276 kPa

2000 1500 1000 500 0 0 5 10 15 20 25 Axial Strain (%) 30

Va = 0% a = 10% Dia. Increase 8%

Va = 0% a = 30% Dia. Increase 20%

Figure 5. Post peak stress-strain curves and specimen shape after testing for SMA.

3.3 Magnitude of Failure Strains and Confinement Effect

Figure 6 summarizes the measured axial failure strains for the studied mixtures for different air voids content and confinement levels. Figure 6(a) shows that failure strains for the DGM ranged from 2 to 8%. The low air voids content mixtures failed approximately 2 to 3% strain, while the failure strain increased with increasing confinement. Figure 6(b) shows that the SMA mixture has slightly higher failure strains ranging from 3 to 21%. It can be speculated that in SMA specimens with 12% air voids at highest confinement level, the large axial failure strain (above 10%) indicates behavioral differences. By examining stress-strain curves from Figures 4(c) and 4(d) it appears like these two specimens were densifying instead of dilating during testing.

3 = 0 kPa

3 = 276 kPa

3 = 138 kPa

3 = 414 kPa

3 = 0 kPa

3 = 276 kPa

3 = 138 kPa

3 = 414 kPa

25

Peak Strain (%)

25

Peak Strain (%)

0 5 10 Air Void Content (%) 15

20 15 10 5 0

20 15 10 5 0 0 5 10 Air Void Content (%) 15

(a)

(b)

Figure 6. Failure Strains at various air voids content levels.

4. ANALYSIS USING MOHR-COULOMB FAILURE THEORY

The triaxial strength test has traditionally been used in soil property characterization to obtain the shear strength of the soil. The Mohr-Coulomb failure theory, Equation (1), describes the relationship between shear strength () and applied normal stress () using two material parameters, cohesion (c) and friction angle ().

= c + tan

Unconfined compressive strength (qu) can be determined from Equation (2).

(Eq.1)

Paper 027

8th CONFERENCE ON ASPHALT PAVEMENTS FOR SOUTHERN AFRICA

qu =

2c cos 1 - sin

(Eq.2)

The triaxial test results are tabulated in Table 2. The table shows the measured deviatoric stress (d) obtained at each confinement level (3). The peak stress was obtained by examining the graphs shown in Figures 2 and 4. For high air voids content mixtures the peak stress was obtained at the highest point regardless of axial strain at that point. Table 2 also shows the computed cohesion (c) and internal friction angle () values for each level of air voids content. Cohesion and friction were estimated using test results from different confinement levels to obtain at least three points in the failure line. The test data was plotted to a p-q stress plane, Figure 7, then, using linear regression the intercept (a0) and slope (a1) of the regression line were computed. The p-q stress plane was formed by defining p and q by Equations (3) and (4) [20], and cohesion and friction angle were computed using Equations (5) and (6).

Table 2. Triaxial shear strength test results.

Confinement 3 (kPa) 0 138 276 414 0 138 276 414 0 138 276 414 0 138 276 414

Measured Va (%) 13.0 13.0 13.0 13.7 8.1 8.3 8.3 8.8 3.7 3.8 4.0 4.1 0.0 0.0 0.0 0.0

DGM Deviatoric Stress, d (kPa) 576 1167 1504 1733 947 1557 2017 2660 1446 1828 2384 2524 1792 * 2602 *

c (kPa)/ (Deg.) 168 / 35.9

Measured Va (%) 12.1 12.0 12.1 12.2 7.8 7.6 8.4 7.8 3.2 3.4 3.4 3.3 0.0 0.2 0.0 0.1

202 / 43.0

374 / 35.8

452 / 36.5

SMA Deviatoric Stress, d (kPa) 654 1206 1652 2302 1094 1594 1784 2247 1375 1755 2162 2525 1648 2253 2462 3089

c (kPa)/ (Deg.) 145 / 41.5

271 / 37.9

353 / 35.7

400 / 38.8

*) Data point not reported due to testing problems

+3 p= 1 2

(Eq. 3)

-3 q= 1 2

= sin -1 (a1 )

0 cos a

(Eq.4)

(Eq.5) (Eq.6)

c=

Paper 027

8th CONFERENCE ON ASPHALT PAVEMENTS FOR SOUTHERN AFRICA

Figure 7. P-q stress plane to obtain cohesion (c) and friction angle ().

Figure 8 shows the cohesion and friction angle as a function of air voids content. Figure 8(a) shows that the cohesion decreased as the air voids content increased. This is in agreement with literature and confirms that the binder stiffness is the major source of cohesion in the asphalt mixtures. The trend for the internal friction angle is not clear as Figure 8(b) shows. Both mixtures seem to follow a similar trend for cohesion but the behavior of the friction angle seems to deviate slightly. Mixture cohesion at refusal density was 2.7 times higher than at 13% air voids content for the DGM and 2.8 for the SMA. The maximum friction angle was 1.2 times higher than the minimum friction angle for the DGM and 1.6 for the SMA. It is evident that cohesion is more influenced by the air voids content, although it is not known how sensitive the asphalt mixture performance is for the range of variation of these parameters.

DGM 500 400 c (kPa) 300 200 100 0

(a)

SMA

50 40 (deg.) 30 20 10 0 0 2

DGM

SMA

0

2

4

6

8 10 12 14

Air Void Content (%)

(b)

4

6

8

10

12 14

Air Void Content (%)

Figure 8. (a) Cohesion and (b) Internal friction angle vs. air voids content.

To investigate the effect of confinement on mixture strength, Figure 9 shows the unconfined compressive strength (qu) and shear strength () plotted as a function of air voids content. Shear strength was computed using Equation (1) with 300 kPa and 700 kPa normal stresses. The 300 kPa normal stress represents hypothetical pavement stress at the edge of the tire at 75-mm deep in the pavement. The 700-kPa normal stress represents stress state at 50 mm deep underneath the tire. These calculations were done analyzing pavement structure of 150 mm asphalt mixture (E = 690 MPa) over 150 mm of granular base (E = 165 MPa) by Layered Elastic Analysis. The subgrade modulus was 55 MPa, and in both situations a 690-kPa-tire pressure was used to compute stresses.

Paper 027

8th CONFERENCE ON ASPHALT PAVEMENTS FOR SOUTHERN AFRICA

Figure 9. (qu) and () vs. air voids content.

The DGM with low air voids has higher cohesion and unconfined strength than the SMA although the SMA mixture has stiffer binder. At around 6% air voids content the order of the mixtures reversed. In contrast to the unconfined compressive strength, the shear strength seemed to be less sensitive for the change in air voids content, as Figure 10 shows. Also, mixtures seem to deviate less from each other. With normal stress of 300-kPa the shear strength and unconfined strength seemed to merge to the same magnitude at around 10.5% air voids content. This indicates that when mix has high air voids content the shear resistance is rising mostly from cohesion because aggregate interlock is reduced or prevented in the mixture. Figure 10 shows the ratio of the confined and unconfined peak deviatoric stresses. In Figure 7, (dc) is the peak deviatoric stress at confined state and (du) is the peak deviatoric stress at unconfined state. Figure 9(a) shows that the shear strength of the DGM mix is more dependent on the level of confinement as the air voids in the mix increases compared to the SMA mixture. Figure 9(b) shows that in the SMA mixture the ratio is almost insensitive to the confinement up to 8% air voids content. This suggests that the aggregate skeleton responds differently to the applied load in these two different mix types, as expected.

Conf. = 138 kPa Conf. = 414 kPa Conf. = 276 kPa Conf. = 138 kPa Conf. = 414 kPa Conf. = 276 kPa

dc / du

2.5 2.0 1.5 1.0 0 5 10 Air Void Content (%) 15

dc / du

4.0 3.5 3.0

DGM

4.0 3.5 3.0 2.5 2.0 1.5 1.0 0

SMA

(a)

(b)

5 10 Air Void Content (%)

15

Figure 10. Shear strength dependency on confinement and air voids content.

Paper 027

8th CONFERENCE ON ASPHALT PAVEMENTS FOR SOUTHERN AFRICA

5. ANALYSIS USING CRITICAL STATE SOIL MECHANICS THEORY

Test results were also analyzed using Critical State Soil Mechanics theory (CSSM), which has been developed originally for geotechnical engineering. The CSSM theory states that the soil body, when loaded continuously, will be ultimately sheared at constant shear stress, constant effective stress, and constant volume (void ratio). In field and laboratory testing, soil is loaded from an initial state and will reach a critical state, which is related to the failure of soil body. At critical state, the soil is sheared at constant shear stress, constant effective stress, and constant volume (void ratio) and there is a unique relationship between the shear strength, the normal stress and the void ratio. The essence of critical state is that, during shearing, all soils will ultimately reach their critical states and the ultimate or critical states are independent of the initial states [21]. Instead of adopting the Mohr-Coulomb's equation to obtain (c) and (), critical state soil mechanics characterizes peak strength, denoted by peak friction angle, p, of soil as the sum of critical state angle (c) and dilatation angle (). Bolton [22] proposed a relationship between these two by introducing the dilatancy index (IR) for triaxial conditions, where IR is actually an index of (), Equation (7).

p = c + 3I R

(Eq.7)

For the triaxial strength test, peak and critical state friction angle can be obtained using Equations (8) and (9). Np is the flow number for the peak state and Nc is the flow number for the critical state. They are determined from the ratio of major principal stress (1) and minor principal stress (3) at the corresponding states. The subscripts p and c in Equations (8) and (9) denote the peak and the critical states, respectively.

tan 2 (45o +

p

2

)=

1 p 3p

= Np

(Eq.8)

tan 2 (45o + c ) = 1c = Nc 2 3c

(Eq.9)

The loose soil is said to be on the wet side of the critical state because it must densify and shear to reach the critical state. The dense soil is said to be on the dry side because after some initial densification it dilates to reach the critical state. Figure 11 illustrates this behavior. The rationale of applying the CSSM in this study is that asphalt mix is expected to behave like cemented sand at high service temperatures when the cohesive force from asphalt is reduced and the interlock among aggregate particles becomes the major source of shear resistance. Based on O'Rourke, and Crespo [23] the critical state of soil can be determined by examining the stress-strain curve of the soil; when the stress-strain curve becomes flattened with the increased axial strain, the critical state has been reached. By examining stress-strain curves for both mixtures it is evident that only few specimens reached a constant shear stress with increasing axial strain. This indicates that for most of the specimens the critical state was not reached during testing. Also, the stress-strain curves suggest that almost all specimens dilated during testing and only a few specimens densified being on the "wet" side of the critical state. This behavior is due to the very high cohesion caused by the binder in the mixtures.

Paper 027

8th CONFERENCE ON ASPHALT PAVEMENTS FOR SOUTHERN AFRICA

Figure 11. Soil behavior in wet and dry condition.

This suggests that the laboratory testing with the loading ram rate of 50-mm/min cannot capture the in-situ densification behavior of asphalt mixtures under traffic loading. Also, it seems to be evident that tested at 50 mm/min ram rate, asphalt mixtures do not seem to have critical state. The large bulging of specimens at high axial strain due to the additional restraint caused by membrane around the specimen may distort these findings somewhat. However, the overall behavior of the mix is evident and additional investigation of existence of critical state must be conducted using slower loading rate to reduce the cohesion coming from the asphalt binder. Although the critical state was not confirmed, the dilatation behavior of mixtures was studied using the above-discussed concept of obtaining dilatation index (IR). A state corresponding to the 10% axial strain was decided to use as a pseudo critical state for comparison purposes.

Va = 0%

2500

Np

Va = 4%

Va = 8%

Va = 12%

2000

d, kPa

3 = 138 kPa

1500 1000 500 0 0 5 10 15 20 25 Axial Strain (%) 30 35

Nc

Figure 12. Determination of peak and critical state flow number from stress-strain curve.

Figure 12 shows the procedure used to obtain the flow number for peak and critical state to compute the friction angles from stress-strain curves. Figure 13 shows the computed dilatancy index values for the DGM and SMA mixtures for 138-kPa confinement levels. The DGM seems to have more dilatancy potential than the SMA mixture. Additionally, the dense specimens

Paper 027

8th CONFERENCE ON ASPHALT PAVEMENTS FOR SOUTHERN AFRICA

possess higher dilatancy potential than the more loose specimens, as expected. Specimens with air voids content of 12 to 13% show very limited signs of dilatation and the specimen failure is envisioned as results of pure sliding or rearrangement of aggregate particles as opposed to climb-past phenomenon in dense specimens resulting in lower shear resistance. Figure 13(a) shows the computed dilatancy index for the DGM and SMA mixtures as a function of voids filled with asphalt (VFA). The SMA mixture is less sensitive for dilative behavior and the dilatation index is basically constant up to 80% VFA, which represents about 3.5% air voids content. This may suggest that the SMA mixtures must be compacted at least 4% air voids content in order to have the best aggregate interlock in the aggregate skeleton. On the other had, this also suggests that the strength of SMA mixture is not so sensitive for the changes of in-situ air voids content compared to the DGM. Figure 13(b) shows the shear strength of mixtures as a function of dilatation index. Shear strength was computed for 300 and 700-kPa normal stress. At lower normal stress the SMA mixture had higher shear strength with the same dilatation index value. Also, with higher normal stress the SMA retained higher shear strength as the dilatation index increased. This suggests that the SMA mixture is less prone to rutting by shear and densification compared to the DGM. In order for traffic to densify mixtures the rearrangement of aggregate structure must take place by coupled action of volumetric and shear straining. Based on these findings the DGM seems to be more prone to dilatation and shear compared to the SMA mixture.

DGM 16 Dilatancy Index

SMA 1100 1000 900 800 700 600 500 400 300

DGM @700kPa DGM @300kPa

SMA @700kPa SMA @300kPa

8 4 0

(a)

40

50 60

70

80

90 100

Shear Sress (kPa)

12

VFA (%)

(b)

0

4

8 12 Dilatancy Index

16

Figure 13. Dilatancy index and shear stress vs. VFA.

6. SUMMARY AND CONCLUSIONS

Two asphalt mixtures, a dense graded mix (DGM) and a Stone Mastics Asphalt (SMA), were investigated using triaxial shear strength testing at 50 mm/min ram rate loading at 55°C. The objective of the study was to investigate the effect of air voids content on the mix strength properties by varying air voids content from 0 to 13%. The triaxial testing was conducted using 0, 138, 276 and 414-kPa confinements. Two different theories, the Mohr-Coulomb failure theory and Critical State Soil Mechanics theory (CSSM) were used to analyze the test data. Analysis using Mohr-Coulomb failure theory shows that both the DGM and SMA mixture had highest cohesion, unconfined compressive strength (qu) and shear strength at refusal density, i.e., at zero air voids content. The DGM had higher cohesion and (qu) below 4% air voids content than the SMA but both had similar shear strength properties. No clear trend for the friction angle as a function of air voids content was obtained. Analysis however suggests that the high air voids content (12 to 13%) prevents mixtures to develop aggregate interlock and therefore shear resistance.

Paper 027

8th CONFERENCE ON ASPHALT PAVEMENTS FOR SOUTHERN AFRICA

Overall, the existence of critical state for asphalt mixtures could not be confirmed or rejected using CSSM. There are several reasons for that; first the loading rate used in the research was too high to eliminate cohesion in the mixtures, which prevented mixture densification. Secondly, the accurate volume changes were not measured due to the preliminary nature of the research. Also, the membrane around the specimen restricted movements in the aggregate skeleton causing some unknown error for the post peak test results. However, by using 10% axial strain as a pseudo critical state, a dilatation index was computed which suggests that the shear strength of the SMA mixture is less sensitive to the varying air voids content and changes in the VFA than the DGM. This suggests that the SMA mixture is more rut resistant compared to the DGM.

7. ACKNOWLEDGEMENTS

The financial support from Purdue Research Foundation (PRF) is highly appreciated. Also, Mr. Mark Baker from Purdue Asphalt Laboratory provided valuable assistance in performing the testing.

8. REFERENCES

Bonnaure, F., Guest, G., Gravois, A., & Uge, P. (1977). A New Method of Predicting the Stiffness of Asphalt Paving Mixtures. Journal of the Association of Asphalt Paving Technologist, Vol. 46. Andrei, D., Witczak, M., W., & Mirza, M. W. (1999). Development of Revised Predictive Model for the Dynamic (Complex) Modulus of Asphalt Mixtures. Development of the 2002 Guide for the Design of New and Rehabilitated Pavement Structures, NCHRP 1-37A. Interim Team Technical Report. Department of Civil Engineering, University of Maryland of College Park, MD. Christensen D.W., Pellinen, T. and Bonaquist, R.F. (2002) Hirsh Model for Estimating the Modulus of Asphalt Concrete. Paper prepared for presentation at the Annual Meeting of the Highway Research Board. Pellinen, T., and Witczak, M. W. (1998). Assessment of the Relationship between Advanced Material Characterization Parameters and the Volumetric Properties of a MDOT Superpave Mix. Submitted to Maryland Department of Transportation State Highway Administration. Department of Civil Engineering, University of Maryland at College Park, MD. Pellinen, T. (2003). The Effect of Volumetric Properties on Mechanic Behavior of Asphalt Mixtures. Transportation Research Board 2003 Annual Meeting, Washington, D.C. Christensen D.W., Bonaquist R., and Jack D. (2000) Evaluation of Triaxial Strength as Simple Test for Asphalt Concrete Rut Resistance. Final Report, Pennsylvania Transportation Institute, the Pennsylvania State University, University Park, PA. Huang, E.Y. (1970) A Study of Strength Characteristics of Asphalt-Aggregate Mixtures as Affected by the Geometric Characteristics and Gradation of Aggregates. Proceeding of the Association of Asphalt Paving Technologists. Vol. 39, pp. 98-133. Huschek, S. (1985) The Deformation Behavior of Asphaltic Concrete Under Triaxial Compression. Proceeding of the Association of Asphalt Paving Technologists. Vol. 54, pp. 407-426. Krutz, N.Z. and Sebaaly, P.E. (1993) The Effect of Aggregate Gradation on Deformation of Asphalt Concrete. Proceeding of the Association of Asphalt Paving Technologists. Vol. 62, pp. 450-473. Stiady, J.L., Hand, A.J., Noureldin, A.S., Hua, J., and White, T.D. (2002) Validation of SHRP Mixture Specifications Using Accelerated Testing. National Pooled Fund Research No. 176, Final Draft Report, INDOT.

Paper 027

8th CONFERENCE ON ASPHALT PAVEMENTS FOR SOUTHERN AFRICA

McLeod, N. W. (1948). The Stability of Granular and Cohesive Materials in Triaxial Compression. Journal of the Association of Asphalt Paving Technologist, Vol. 17, pp.184 -215. Mcleod N.W. (1952) Rational Design of Bituminous Paving Mixtures with Curved Mohr-Coulomb Envelopes. Proceeding of the Association of Asphalt Paving Technologists. Vol. 21, pp. 349-437. Nijboer, J.W. (1948) Plasticity as a Factor in the Design of Dense Bituminous Road Carpets. New York: Elsevier Publishing Co., Inc., pp 148. Hewitt, W.H. (July, 1964) Analysis of Various Flexible Paving Mixtures by Theoretical Design Procedure Based on Shear Strength. Paper prepared for presentation at the Annual Meeting of the Highway Research Board. The Marshall Method for the Design of Bituminous Paving Mixtures. 3rd rev., Marshall Consulting and Testing Laboratory, Jackson, MS 1949. InstroTek, Inc. (2002) CoreLok Operator's Guide. InstroTek, Incorporated, Raleigh, NC. Witczak, M.W., Bonaquist, R., Von Quintus, H., and Kaloush K., (2000) Specimen Geometry and Aggregate Size Effects in Uniaxial Compression and Constant Height Shear Test. Preprint of Proceedings of the Association of Asphalt Paving Technologists, Vol 69. Schnaid, F., Prietto, P.D.M., and Consoli, N.C. (2001) Characterization of Cemented Sand in Triaxial Compression. Journal of Geotechnical and Geoenvironmental Engineering, Vol 127, No. 10, pp. 857-868. Scarpas, A. and Blaauwendraad, J. (1998) Experimental Calibration of a Constitutive Model for Asphalt Concrete. Proc. Euro-C Conf. on the Computational Modeling of Concrete Structures, Badgastein, Austria, pp. 193-202. Bowles, J.E. (1979) Physical and Geotechnical Properties of Soils. New York: McGraw-Hill Book Co., pp. 478. Atkinson, J. (1993). An Introduction to the Mechanics of Soils and Foundations. McGraw-Hill International Series in Civil Engineering, London, UK. Bolton, M.D. (1986) The Strength and Dilatancy of Sands. Geotechnique, London, 36(1), pp. 65-78. O'Rourke, T.D., and Crespo, E. (1988) Geotechnical Properties of Cemented Volcanic Soil. Journal of Geotechnical Engineering, Vol. 114, No. 10, pp.1126-1147.

Paper 027

Information

Characterization of Hot Mix Asphalt with Varying Air Voids Content Using Triaxial Shear Strength Test

15 pages

Report File (DMCA)

Our content is added by our users. We aim to remove reported files within 1 working day. Please use this link to notify us:

Report this file as copyright or inappropriate

9817


You might also be interested in

BETA
Asphalt Magazine
TABLE OF CONTENTS
EFFECT OF AGGREGATE GRADATION ON PERMANENT DEFORMATION POTENTIAL OF
Mix Design Procedures