Effect of asphalt thin beams mixed with three nominal maximum aggregate sizes in the bending beam rheometer on the prediction of thermal properties of bituminous material

Abstract

The objective of this research is to evaluate an impact of asphalt mixture beams with varying sizes of aggregate in the Bending Beam Rheometer (BBR) for testing/predicting thermal cracking properties of asphalt pavements. The BBR test has following benefits: the equipment is cheaper, it uses smaller specimens, faster conditioning, easier availability for quality control, easier to manage, etc. However some concerns have been raised: some consider that the size of the aggregate may affect the test’s results; the other concern is that such small beams cannot represent the whole properties of the asphalt pavement. To address these criticisms, imaging techniques, statistical analysis, and viscoelastic modeling are used. Asphalt thin beams prepared with three different nominal maximum aggregate size (NMAS) (12.5mm, 9.5 mm, and 4.75 mm) were tested at three different temperatures (-18°C, -24°C, and -30°C). Based on results from statistical analyses and viscoelasticity, the ratio of asphalt binders and voids and stiffness differences among the three NMAS specimens are not significant, meaning that the impact of asphalt thin beams prepared with the three NMAS on the prediction of thermal cracking is minimal and can be neglected.

Keywords

bending beam rheometer / thin beams / thermal cracking

Background

In cold weather climates, low temperature cracking of asphalt pavements is one of the most prevalent distresses. Currently, there is no quality control (QC) tests that were adopted by highway agencies for predicting thermal stresses due to the fact current standard methods for testing asphalt mixture beams (e.g., indirect tensile test, Thermal Stress Restraint Specimen Test) might not be adequate and practical during day-to-day operations and construction. Many researchers have worked on seeking possible solutions and material testing methods to address the above issue. One of solutions is the application of asphalt mixture beams in a bending beam rheometer (BBR). It is well known that the BBR test is exclusively used for testing asphalt binders in a laboratory. However, based on previous research works implemented by researchers at the University of Minnesota and the University of Utah, the BBR could be used by highway agencies as a feasible testing method for regular control of asphalt mixtures design and QC during construction of asphalt pavements. Their work indicated that using the BBR for measurement of the low temperature properties of asphalt mixtures has been widely studied and supported [ 1– 3]. Asphalt industry as a whole has been skeptical in using thin asphalt beams in the BBR to evaluate the global properties of asphalt mixtures. One of concerns is due to small aggregate sizes used in a BBR thin beam (112 mm length × 12.7 mm width × 6.35 mm thickness) (Fig. 1) and that such small specimens cannot represent the global properties of asphalt mixtures. Clendennen and Romero [ 4] examined the effect of representative volume element (RVE) for testing in a bending beam rheometer (BBR) using Bartlet method (a statistical theorem). To the extent of their effort, the paper is further presented to evaluate the effect of mixtures prepared with three nominal maximum aggregate sizes (NMAS) (4.75mm, 9.5mm, and 12.5mm) on the performance of asphalt pavements using imaging technology, statistical methods, and viscoelasticity. The use of BBR in prediction of thermal properties of asphalt mixtures has received increasing attention in the asphalt industry. This paper is presented in support of the above research and provide better understanding of the BBR when used in a laboratory and in the field.

Material preparation

The mixture formulation used in this research project was Hot mix asphalt concrete (HMA). he 12 mm Nominal maximum aggregate size (NMAS), which is defined as one sieve larger than the first sieve to retain more than 10 percent of the material (Roberts, Kandhal, Brown, & Lee, 1996), was selected as the standard mix as this is a typical aggregate size in many cold regions in the US such as Utah, Northern Arizona, Colorado, etc. Another two NMAS (9.5 mm and 4.75 mm) were chosen for purposes of RVE evaluation and mechanical response in comparison with the standard 12.5 mm NMAS. Table 1 shows the mix design for the three NMAS groups. The design was provided by the Utah Department of Transportation in an attempt to evaluate the effect of the three NMAS mixture beams on the prediction of low temperature cracking.

Based on these three NMAS mix design, asphalt samples were compacted using a Superpave Gyratory Compactor (SGC) and then further trimmed two sides off of each sample using a lapidary saw. To evaluate the distributions of RVE on the three NMAS samples, an experiment was implemented using imaging techniques to isolate asphalt binder and air voids (named ABAV) from a specimen and compare the ratio of ABAV occupied in the entire specimen.

Imaging process and statistical analysis

Image analysis method

Two sets of asphalt specimen from each NMAS mix design were compacted with a dimension of 150mm in diameter and 110 mm in height. Specimens were cut both sides off to a rectangular size (110 mm by 127 mm) for imaging analysis. The images of cross section for the three NMAS compacted specimens were obtained from a scanner and were shown in Figs. 2(a) –(c). After image conversion processes using Photoshop, the distributions of ABAV in each NMAS specimen were shown in Figs. 3(a)–(c). Note that this is not a 100% accurate conversion due to the fact that pixels get lost during a conversion of an image. Also there were pixels (especially around the image borders) that had to be omitted because they didn't depict an accurate reading of the asphalt and during the conversion to black and white just became artifacts. However, consider image borders to be appeared for each image, ratio of ABAV in each specimen could be relevantly comparable. The impact of pixels along borders on the conversion results can be neglected. The ratios of ABAV occupied in each specimen were depicted in Table 2.

The objective of the image conversion processes is to evaluate the stability of ratios of ABAV occupied in specimens when converting images into black and white. The major concern during the image conversion is that pixels in black (asphalt binder and air voids) could be miscalculated due to the fact that some of cross sections of coarse aggregates appear to be colored in dark gray which might lead to an increase to the total pixels in black during conversions. Based on the observation in Table 2, it is believed that the impact of gray color of coarse aggregates on pixels of ABAV is insignificant thus considering the image conversions of NMAS specimens to be valid for the calculation of the ratio of ABAV in six NMAS specimens.

For this analysis the asphalt pucks were cut into blocks where the aggregate of the mixture was visible. The images were scanned to be evaluated digitally. A scale factor was used for each NMAS to have equivalent samples and compare them. If the result of counting asphalt binder and air voids is statistically equivalent we can assume that there is no significantly difference among the three NMAS.

The NMAS of 12.5 mm (0.5 in) was used as a reference scale. The others two NMAS were scaled up multiplying by 0.5 in/0.25 in= 200% for NMAS of 4.75 mm (0.25 in), and 0.5 in/0.375 in= 167% for NMAS of 9.5 mm (0.375 in). Figure 4 shows an example of scaled 9.5 NMAS specimen. Two sets of specimens for each NMAS were produced in a laboratory. After these specimens were scanned and save as a digital file, an imaging process was implemented to obtain a number of pixels for each specimen. The number of pixels represent the volume of asphalt binder and voids occupied within a specimen. Figure 4 shows close-up of a specimen before and after imaging process. The purpose of the imaging analysis using three scale factors is to assess even if the ABAV varies from the three NMAS, at low temperatures, the binder tends to be stiffer so its stiffness would merge with the aggregate, making the effect of aggregate sizes on the global properties of the thin beam was minimal.

Statistical analysis

As there were two sets of scaled images available for analysis, one of them was randomly selected for statistical analysis. A one way analysis of variance (ANOVA) was performed to validate the hypothesis that the ABAV difference between three NMAS specimens (12.5NMAS, scaled 9.5NMAS, and scaled 4.25NMAS) is not significant. The result is depicted in Table 3. Based on the calculated p-value, we fail to reject the hypothesis which means the ABAV different among the three NMAS specimens is not significant.

The ANOVA analysis indicates that after scaled up of 9.5 NMAS and 4.75 NMAS specimens, the ABAV values among the three NMAS specimens is statistically close suggesting that the effect of the three NMAS mixture beams on the ABAV of asphalt mixtures prepared with three aggregate sizes can be neglected.

Viscoelastic analysis for three NMAS specimens

Through imaging process and statistical analysis, it is known that the effect of asphalt thin beams prepared with three NMAS does not have significant affect the global properties of asphalt mixtures. The next step is to run viscoelastic modeling from the three NMAS specimens and compare their mechanical responses.

Sample preparation and testing

Creep compliance data were obtained from the BBR test to characterize the low temperature properties of asphalt mixtures beams. Mixture beams varied with the nominal maximum aggregate size (NMAS) (12.5mm, 9.5 mm and 4.75 mm). All SGC specimens were trimmed into several thin beam with dimensions suitable for the BBR test (12.7 mm x 6.35 mm x 127 mm) (width x thickness x length).

The test was run following the AASHTO T313 standard [ 5]. Before running the test, the beams were conditioned in a bath for an hour at three temperatures (-18°C, -24°C and -30°C). The initial load is 35±10 mN, but it was increased to 450-g (4413±50 mN) to obtain measurable deflections without change in the cell calibration procedure [ 6]. The computer program collects creep data based on 0.5 s increment which were the base to determine afterward the creep compliance, and relaxation modulus of asphalt mixtures through a Laplace transformation process.

Viscoelastic analysis

The linear viscoelastic (LVE) modeling analysis is used to evaluate if the relaxation moduli of the three NMAS specimens obtained from the BBR test are similar and all of them can be used to estimate the low temperature properties of asphalt concrete mixtures.

A power law function was used to model the responses of creep compliance, using previous research recommendations [ 6]:

(1) D(t)=D_{\mathrm{0}}+D_{\mathrm{1}}{t}^n(\mathrm{Power}\mathrm{law}),

where D(t) is creep compliance at reduced time t; D₀, and D₁ and n are power function parameters.

Creep compliance is defined as a time-dependent ratio of strain e(t) when subjected to a constant stress and relaxation modulus can be referred to as a time-dependent ratio of stress s(t) when subjected to a constant strain. While there are many LVE models available to perform relaxation calculation, the power law function is one of commonly used methods for the LVE analysis purposes. A time-temperature superposition principle associated with a pre-smoothing technique provided by recent studies [ 7] were used to produce a fitted master curve representing the linear viscoelastic responses of the experimental data. Using the fundamental principle of nonlinear regression (minimizing the sum of squared errors between the raw data and fitted values), the power law parameters (D₀, D₁ and n) were calculated.

It is known that the relaxation modulus E(t) can be determined by performing Laplace transformation of creep compliance function D(t) (Eq. (1)). The details can be found from Ho and Romero [ 8]. The relaxation modulus E(t), is therefore expressed:

(2) E(t)={\displaystyle \frac{\mathrm{1}}{D_{\mathrm{0}}+D_{\mathrm{1}}\Gamma (n+\mathrm{1}){(\mathrm{1.786})}^n}},

where \Gamma is defined as a gamma function.

Results of viscoelastic analysis

The creep compliance data of the asphalt concrete mixtures for each NMAS at the three different temperatures is drawn against a logarithm time scale. By using the Time-Temperature Superposition principle, the three individual creep compliance curve were shifted to partially overlapped together as shown in Fig. 5. The master curve was developed to fit the raw data of the creep compliance (Fig. 5) using the minimum sum squired error method. The three power function parameters were therefore determined and are shown in Table 4.

Inserting parameters from Table 4 in Eq. (2), the three relaxation moduli were yielded as shown in Fig. 6. It is clearly noticed that while the three relaxation modulus curves are not identical but they show close agreement. The LVE result provides promising indication to the prediction of thermal cracking among the three NMAS specimens. As previously mentioned, there is a concern on the use of small thin beams in representation to the global properties of asphalt mixtures. However, like many previous researchers [ 1, 2, 9] pointed out when the temperature drops, the aggregate and binder within the asphalt matric tend to merge together thus that the use of small thin beams in the BBR can be applied in the prediction of thermal cracking of asphalt mixtures. This paper is presented to support the above statement and concludes the effect of asphalt thin beams prepared with the three NMAS specimens in the BBR is minimal when predicting the thermal cracking of asphalt mixtures.

Conclusions

This paper discusses the effect of asphalt thin beams prepared with the three NMAS in the BBR on the prediction of thermal cracking of asphalt mixtures and presents three methods using imaging technology, statistical analysis, and linear viscoelastic modeling. The following conclusions are drawn based on the above analyses.

1) A series of imaging techniques were used to compare the ratio of ABAV of the three NMAS specimens. The p-value of a one way ANOVA test is greater than 0.05, so the ABAV difference among the three NMAS specimens (12.5 mm, 9.5mm, and 4.75mm) is not significant.

2) Based on the result viscoelastic modeling, the modulus curves of three NMAS specimens show good agreement. The effect of aggregate sizes on low temperature cracking is not significant, thus the three NMA size analyzed (4.75 mm, 9.5 mm and 12.50 mm) have close viscoelastic response at low temperatures.

3) All test data conclude that the impact of asphalt beams prepared with the three NMAS in the BBR on the prediction of thermal cracking of asphalt mixtures is minimal and can be neglected.

References

[1] Zofka A, Marasteanu M, Li X, Clyne T, McGraw J. Simple method to obtain asphalt binders low temperature properties from asphalt mixtures properties. Association of Asphalt Paving Technologists, 2005, 74: 255–282

[2] Velasques R M. Effect of beam size on the creep stiffness of asphalt mixtures and low temperatures. Advanced Testing and Characteriztion of Bituminous Material, 2009

[3] Ho C H, Romero P. Using asphalt mixture beams in the bending beam rheometer. Experimental and Numerical Approach. Road Materials and Pavement Design, 2011, 12(2): 293–314

[4] Clendennen C R, Romero P. Evaluating the representative volume element of asphalt concrete mixture beams for the testing in the bending beam rheometer. Multi-Scale Modeling and Characterization of Infrastructure Materials: Proceedings of the International RILEM Symposium. Stockholm, Sweden, 2013, 13–30

[5] American Association of State Highway and Transportation Officials. Determining the Flexural Creep Stiffness of Asphalt Binder Using the Bending Beam Rheometer (BBR). In: Standard Specifications for Transportation Materials and Methods of Sampling and Testing T313. 29 ed. 2009

[6] Christensen D W, Bonaquist R. Evalutation of Indirect Tensile Test (IDT) Procedures for Low-Temperature Performance of Hot Mix Asphalt. Transportation Research Board, National Cooperative Highway Research Program. Report 530, Washington, DC, 2004 <PublisherName Language="chs"/>

[7] Park S, Kim Y. Fitting prony-series viscoelastic models with power-law presmoothing. Journal of Materials in Civil Engineering, 2001, 13(1): 26–32

[8] Ho C H, Romero P. Using linear viscoelastic modeling to evaluate the low temperature properties of asphalt mixtures prepared with aggregates of different sizes. Journal of Advances in Civil Engineering Materials, the American Society for Testing and Materials (ASTM). International, 2013, 2(1): 122–139

[9] Zofka A M. Investigation of Asphalt Concrete Creep Behavior Using 3-Point Bending Test. University of Minnesota. Minnesota: ProQuest, 2007

RIGHTS & PERMISSIONS

Higher Education Press and Springer-Verlag Berlin Heidelberg