- Email: [email protected]
Relating asphalt binder elastic recovery properties to HMA crack modeling and fatigue life prediction
Construction and Building Materials 111 (2016) 644–651
Contents lists available at ScienceDirect
Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat
Relating asphalt binder elastic recovery properties to HMA crack modeling and fatigue life prediction Jun Zhang a,⇑, Geoffrey S. Simate b, Sang Ick Lee c, Sheng Hu c, Lubinda F. Walubita c,d,e a
Department of Civil Engineering, Texas A&M University, College Station, TX 77843, USA University of the Witwatersrand, Johannesburg, P/Bag 3, Wits 2050, South Africa c Texas A&M Transportation Institute (TTI) – The Texas A&M University System, College Station, TX 77843, USA d Iliso (Z) Consulting Engineers Ltd, Lusaka, Zambia e Transportation Research Center, Wuhan Institute of Technology, Wuhan, China b
h i g h l i g h t s Binder elastic recovery is hypothesized to play a significant role in the fatigue cracking resistance. Two standardized tests are used to characterize binder elastic recovery properties. Two laboratory tests are used to characterize HMA properties of stiffness and fracture. HMA with high binder elastic recovery properties (P59%) has a long M-E predicted fatigue life. No clear trend is observed in predicted fatigue life for HMA with low elastic recovery (<59%).
a r t i c l e
i n f o
Article history: Received 11 November 2015 Received in revised form 4 February 2016 Accepted 23 February 2016 Available online 21 March 2016 Keywords: Asphalt binder Elastic recovery MSCR HMA Cracking Fracture properties Fatigue life Mechanistic-empirical (M-E) TxME
a b s t r a c t Fatigue cracking is one of the major distresses occurring in hot-mix asphalt (HMA) pavements, which is a consequence of accumulation of damage under repeated load applications and changes in the environmental conditions. HMA is predominantly composed of aggregates and asphalt binder, where the latter plays a significant role in the HMA fatigue performance. The elastic recovery of asphalt binders is one of the characteristic properties that are hypothesized to play a significant role in the fatigue cracking resistance of HMA pavements. In this study, the relationships between the asphalt binder elastic recovery properties and the HMA fatigue performance were comparatively investigated. Two asphalt binder tests, namely the elastic recovery (ER) and multiple stress creep recovery (MSCR) tests were conducted to comparatively characterize the asphalt binder elastic recovery properties in the laboratory, and their results were then correlated to the predicted HMA fatigue life based on the TxME modeling software, which was subsequently validated with other laboratory test results and field performance data. Eleven typical Texas HMA mixes collected from different construction sites were used for the study. Overall, the results indicated that HMA mixes with high asphalt binder elastic recovery properties (P59%) exhibited better cracking resistance potential with long predicted fatigue life (>150 months), which was also consistent with other studies and field performance observations. For mixes with low elastic recovery properties (<59%), there is no clear correlation with the predicted fatigue life. Ó 2016 Elsevier Ltd. All rights reserved.
1. Introduction Fatigue cracking is one of the major distresses occurring in hotmix asphalt (HMA) pavements, which is a consequence of accumulation of damage under repeated load applications [1]. HMA is predominantly composed of aggregates and asphalt binder, where the latter plays a significant role in the HMA fatigue performance. ⇑ Corresponding author. E-mail address: [email protected] (J. Zhang). http://dx.doi.org/10.1016/j.conbuildmat.2016.02.175 0950-0618/Ó 2016 Elsevier Ltd. All rights reserved.
Superpave Performance Grade (PG) specification addresses the asphalt binder property related to HMA fatigue performance by measuring a parameter G⁄sin d at intermediate temperatures, where G⁄ and d are the complex shear modulus and phase angle, respectively. However, this parameter was initially and primarily developed for non-modified asphalt binders [2]. With the current wide spread usage of modified asphalt binders, it is now becoming a challenge to effectively relate and quantify the expected HMA mix fatigue performance using the G⁄sin d parameter. Thus, this study was undertaken to investigate alternative asphalt binder
645
J. Zhang et al. / Construction and Building Materials 111 (2016) 644–651
properties relative to the HMA fatigue performance, namely the asphalt binder elastic recovery property. Polymer modified asphalt binders are widely used in HMA pavements, and one of the major improvements in these modified asphalt binders is the elastic recovery property, which is the degree to which an asphalt binder recovers to its original shape after release of the loading application. It is hypothesized that greater propensity to elastic recovery is desirable in HMA pavement to increase the fatigue cracking resistance performance [2]. Few studies have been conducted to investigate the relationship between the asphalt binder elastic recovery property and HMA fatigue characteristics [2]. Based on the aforementioned discussions, this study was initiated to investigate the relationship between the asphalt binder elastic recovery properties and HMA fatigue performance. To achieve this objective, two asphalt binder tests, namely elastic recovery (ER) and multiple stress creep recovery (MSCR); and two HMA performance tests, namely the, Dynamic Modulus (DM) and OT fracture test, were conducted [3–6]. The corresponding laboratory test results were then used to predict HMA fatigue life based on mechanistic-empirical (M-E) modeling with the Texas Mechanistic-Empirical (TxME) Flexible Pavement System software, which was subsequently validated with field performance data. Overall, this study was initiated to address the following three questions: (1) What is the relationship between ER and MSCR results for measured asphalt binder elastic recovery property? (2) What is the relationship between asphalt binder elastic recovery property and asphalt mixture stiffness? (3) What is the relationship between asphalt binder elastic recovery property and HMA fatigue performance? 2. Experimental plan Fig. 1 presents a schematic of the laboratory tests and the experimental research plan that was employed in this study. Two standardized asphalt binder tests, namely the ER and MSCR tests and two HMA performance tests including the DM and OT fracture were conducted. As illustrated in Fig. 1, the DM and OT fracture tests also provided input data for the M-E modeling with the TxME software in addition to characterizing the fundamental HMA properties such as modulus and fracture properties.
3. Laboratory test methods This section presents details of the laboratory tests employed in this study and includes the ER, MSCR, DM and OT fracture tests. 3.1. The ER test The elastic recovery test was performed in accordance with AASHTO T301 using a ductilometer and briquette specimens shown in Fig. 2 [3]. The specimens were conditioned in a water bath at 25 °C for one hour. The test was displacement-controlled and the asphalt binder specimens were pulled apart at a constant speed of 5 cm/min to an elongation of 20 cm, and held for 5 min. Then, a cut was made in the middle of the specimen and allowed them to remain in the ductilometer for recovery. After one hour, the percent elongation recovery was determined using the following equation:
% elongation ¼
20 x 100 20
ð1Þ
where x = final reading in centimeters after bringing the two severed ends of the specimen back together [3].
Fig. 1. Schematic illustration of the laboratory tests and experimental plan.
3.2. The MSCR test The MSCR test is creep and recovery test performed on an asphalt binder sample using the Dynamic Shear Rheometer (DSR) in accordance with AASHTO TP 70-10 [4]. The asphalt binder sample was 25 mm in diameter and 1 mm in thickness. The test was conducted at the high working temperature of the asphalt binder, which was controlled by the use of a water bath in the DSR machine setup. For each cycle, a Haversine shear load was applied to the sample for 1 s, followed by a 9-second rest period. Two stress levels of 0.1 kPa and 3.2 kPa were applied successively, and a total of 10 cycles were conducted for each stress level. A typical MSCR test results with two applied stress levels are shown in Fig. 3. As observed from the figure, the MSCR test characterizes the recovery properties of the asphalt binder under the shear creep load, which is the percent recovery. 3.3. The DM test Unconfined DM testing is an AASHTO standardized test method used for characterizing the stiffness, measured in terms of the dynamic complex modulus (|E⁄|), and visco-elastic properties of HMA mixes (AASHTO TP62-03) [6]. The DM is a stress-controlled test involving application of a repetitive sinusoidal dynamic compressive axial load (stress) to an unconfined HMA specimen over a range of different temperatures (i.e., 10, 4.4, 21.1, 37.8 and 54.4 °C) and loading frequencies (i.e., 0.1, 0.5, 1, 5, 10, 25 Hz) for each temperature. The typical parameter that results from the DM test is the |E⁄|, which was computed as:
jE j ¼
r0 e0
ð2Þ
where r0 is the compressive axial stress and e0 is the corresponding compressive axial resilient strain. For graphical analysis and easy interpretation of the DM test data, |E⁄| master-curves were also generated as a function of the loading frequency using the Pellinen et al.’s [7] time-temperature superposition sigmoidal model shown in Eqs. (3) and (4):
log jE j ¼ d þ
a
ð3Þ
1 þ ebc logðnÞ
logðnÞ ¼ logðf Þ þ logðaT Þ
ð4Þ ⁄
where n is the reduced frequency (Hz), d is the minimum |E | value (MPa), a is the span of |E⁄| values, and b and c are shape parameters. Parameters f and aT are the loading frequency and temperature shift
646
J. Zhang et al. / Construction and Building Materials 111 (2016) 644–651
Fig. 2. Elastic recovery test: (a) briquette specimens; (b) after elongation; (c) cut in the middle of the specimens after elongation.
(0.025 in.) and a loading rate of 10 s per cycle, composed of 5 s loading and 5 s unloading [5]. The test temperature was 25 °C [5]. The ‘‘OT Fracture” test procedure and setup is similar to the standard Tex-248-F illustrated in Fig. 4 except for some minor changes including the reduced opening displacement to 0.425 mm (0.017 in.) from 0.625 mm (0.025 in.) and that the test is run for a maximum of 100 cycles only. However, if the specimen fails in <50 cycles, then the opening displacement should be further reduced to 0.375 mm (0.015 in.) or less until it reaches a minimum of 50 cycles using different specimen sets [8]. The typical output that results from the OT Fracture test are the HMA fracture parameter A and n, which are used as inputs into the TxME software for fatigue life prediction [9]. At a fundamental level, these two parameters, A and n, compositely define the fracture behavior of HMA based on Paris’ law of fracture mechanics as illustrated in Eq. (5) [10].
dc ¼ AðDKÞn dN
Fig. 3. MSCR test: (a) asphalt binder sample loading configuration in the DSR; (b) 10 cycles of creep and recovery at two stress levels of 0.1 kPa and 3.2 kPa.
factor to temperature T, respectively. In this study, 21.1 °C was used as the reference temperature. 3.4. The OT fracture test The OT test is a simple performance test used to characterize the reflection cracking potential of HMA mixes in the laboratory in accordance with Texas specification Tex-248-F [5]. In this test, the HMA specimen was glued onto two metallic plates-one mobile and the other fixed during the testing process. As shown in Fig. 4, there is a 2-mm gap between the two plates to simulate the cracking on an existing old pavement underneath the new constructed pavement. The OT operates in a displacement-controlled loading mode to induce the horizontal movement in the mobile plate to simulate the opening and closing of cracking in the old pavements beneath an HMA overlay. Fig. 4 illustrates the standard Tex-248-F loading configuration consisting of a cyclic triangular displacementcontrolled waveform at a horizontal displacement of 0.625 mm
ð5Þ
where c is the crack length; N is the number of load cycles; dc/dN is the crack speed or rate of crack growth; DK is the change of stress intensity factor (SIF); and A and n are the fracture properties of the material, which are essentially regression coefficients in Eq. (5). Based on Eq. (5), it can be seen that the information required for determining the HMA fracture properties (A and n) are: 1) the crack length c corresponding to a specific number of load cycles N, and 2) the SIF (or DK) corresponding to a specific crack length c. In this study, a software called ‘‘OTA&n” that was developed for fitting the test data onto Paris’ Law model (Eq. (5)) to obtain the regression coefficients of A and n [11]. 4. Materials As listed in Table 1, a total of 11 Texas HMA mixes categorized in five mix types, namely Type B, C, D, F, and CAM, were collected from various field construction sites across Texas and used in this study. Asphalt binders were extracted from these plant-mixed materials and used for ER and MSCR testing. The plant-mix materials were compacted to an air void (AV) 7 ± 1% for HMA performance testing, namely DM and OT fracture [12]. All laboratory testing were conducted on a minimum of three replicate specimens. 5. Results and discussions In this section, the test results for the asphalt binders and HMA mixes are presented, analyzed, and discussed. The relationships between the asphalt binder elastic recovery and HMA fatigue related performance tests were investigated through quantitative
647
J. Zhang et al. / Construction and Building Materials 111 (2016) 644–651
Fig. 4. (a) The Overlay Tester (OT) and HMA specimen Setup; (b) OT loading configuration; (c) example of cracked HMA specimens after OT testing.
Table 1 HMA mix design characteristics. #
Mix label
Mix type
Aggregate gradation
Optimum asphalt content (%)
SHRP continuous grade
Total air voids (%)
1 2 3 4 5 6 7 8 9 10 11
D1 D2 D3 D4 D5 C1 C2 C3 B1 CA1 F1
Type Type Type Type Type Type Type Type Type CAM Type
Dense-graded (12.5 mm NMAS) Dense-graded (12.5 mm NMAS) Dense-graded (12.5 mm NMAS) Dense-graded (12.5 mm NMAS) Dense-graded (12.5 mm NMAS) Dense-graded (16 mm NMAS) Dense-graded (16 mm NMAS) Dense-graded (16 mm NMAS) Coarse-graded (22.4 mm NMAS) Fine-graded (9.5 mm NMAS) Fine-graded (9.5 mm NMAS)
5 5 5 5 5 5 5 5 5 7 6.7
76-29 81-28 70-32 70-32 86-31 78-27 74-30 74-31 83-31 82-34 76-29
7±1 7±1 7±1 7±1 7±1 7±1 7±1 7±1 7±1 7±1 7±1
D D D D D C C C B F
and graphical comparisons. Additionally, comparisons between the HMA fatigue related performance tests and the two asphalt binder recovery are also discussed. 5.1. Relationship between the ER and MSCR tests Table 2 presents the asphalt binder elastic recovery properties measured from the ER and MSCR tests. As noted in Table 2, the ER test was performed at a much lower temperature (25 °C) than MSCR test at 64 °C [3,4].
Fig. 5 shows the correlations of the ER and MSCR test for the asphalt binder elastic recovery. In a previous study conducted by Clopotel et al. (2012), a clear linear relationship was found between the elastic recovery measurements obtained from the ER and MSCR tests [13]. In Fig. 5(a) and (b), it shows a linear trend between the ER and MSCR, but the coefficient of correlation is relatively low (R2 6 0.60), especially for ER versus MSCR-3.2 kPa. Though both ER and MSCR tests measure the asphalt binder elastic recovery property, they are performed at different temperatures that may cause the poor correlation. It can be seen that there exists
648
J. Zhang et al. / Construction and Building Materials 111 (2016) 644–651
Table 2 ER and MSCR test results. Mix label
1 2 3 4 5 6 7 8 9 10 11
D1 D2 D3 D4 D5 C1 C2 C3 B1 CA1 F1
Elastic recovery property (%)
80.0 70.0 60.0 50.0 40.0 30.0 20.0 10.0 0.0
ER (10 °C)
MSCR-0.1 kPa (64 °C)
MSCR-3.2 kPa (64 °C)
59.0 49.0 23.5 23.0 23.0 52.0 29.0 26.8 27.1 70.0 70.0
32.9 42.2 16.0 11.1 42.0 37.5 14.9 16.0 41.4 71.8 61.2
14.9 37.0 4.1 1.2 39.6 24.6 7.4 4.0 42.5 63.4 48.4
y = 0.7539x + 14.591 R² = 0.6002
0.0
20.0
40.0
60.0
80.0
Elastic Recovery_MSCR-0.1kPa (%)
Elastic Recovery_ER (%)
(a) 80.0
5.2. Comparison of asphalt binder elastic recovery property and DM test results
y = 0.5518x + 26.723 R² = 0.3742
70.0 60.0 50.0 40.0 30.0 20.0 10.0 0.0
0.0
20.0
40.0
60.0
80.0
Elastic Recovery_MSCR-3.2kPa (%)
Elastic Recovery_MSCR3.2kPa (%)
(b) 70.0 y = 1.0445x - 10.66 R² = 0.9376
60.0 50.0
Fig. 6 presents the DM (|E⁄|) master curves derived using Eqs. (3) and (4). As can be seen in the figure, CA1 and F1 with the greatest asphalt binder elastic recovery values exhibit lower stiffness when compared to the other mixes. Similarly, D1 has the third largest ER elastic recovery value, and it shows a lower stiffness compared to most of the other mixes. It should be noted that the lower stiffness of CA1 and F1 may be partially attributed to higher asphalt contents as well as higher binder elastic recovery. In addition, DM test results were used as inputs for pavement fatigue performance prediction based on the OT fracture properties in the following section. 5.3. Comparison of the asphalt binder elastic recovery property and predicted pavement fatigue performance based on OT fracture properties
40.0 30.0
In order to evaluate the correlation between the asphalt binder elastic recovery property and pavement fatigue performance of a
20.0 10.0 0.0
0.0
20.0
40.0
60.0
80.0
Elastic Recovery_MSCR-0.1kPa (%)
(c) Fig. 5. Correlations: (a) ER versus MSCR-0.1 kPa; (b) ER versus MSCR-3.2 kPa; (c) MSCR-0.1 versus MSCR-3.2 kPa.
a good linear correlation between the MSCR-0.1 kPa and MSCR3.2 kPa, which may be attributed to the same testing temperature. This also means that for the same temperature, the asphalt binder elastic recovery property at either of two stress levels of the MSCR test can be estimated if the other is known as they have a clear linear relationship (R2 = 0.93).
1.0E+04
Dynamic Modulus (Mpa)
Elastic Recovery_ER (%)
#
Theoretically, the linear correlation exhibited in Fig. 5(c) was expected. This is because, other than the differences in the stress levels (i.e., magnitude), the test loading mode/configuration and temperature were the same since both are MSCR tests. Therefore, the response behavior and ER trend of the different asphalt binders under the two different stress levels (0.1 and 3.2 kPa) of the MSCR test were theoretically expected to be similar as exemplified in Fig. 5(c), with a clear linear relationship and an R2 value of 93%. Other than from the literature data by Clopotel et al. (2012), the relationship between the ER and MSCR tests was not really anticipated and theoretically unknown. For the tested asphalt binders in Fig. 5(a) and (b), the linearity in the correlation is not very strong with the MSCR-0.1 kPa (R2 = 0.60) out-performing the MSCR3.2 kPa (R2 = 0.37). Thus, the MSCR-0.1 kPa would be preferred over the MSCR-3.2 kPa in terms of correlation with the ER test. As previously stated, this poor correlation could be attributed to the differences in the test loading modes (configurations) and test temperatures (i.e., ER at 10 °C and MSCR at 64 °C). By and large, however, the particular asphalt binders (i.e., types, grades, etc.) and the number of asphalt binders or sample replicates (i.e., number of data points) selected for this study could have had an influence on the graphical correlations in Fig. 5(a) and (b). Although it has been concluded from Fig. 5 that the MSCR-0.1 kPa (R2 = 0.60) is superior to the MSCR-3.2 kPa (R2 = 0.37) in terms of correlating with the ER test data, the challenge with the data limitation and relatively poor R2 values suggest additional testing with a wide array of asphalt binders as well as exploring other correlative criteria for the ER-MSCR relationships in future studies.
D1 D2 D3 D4 D5 C1 C2 C3 B1 CA1 F1
1.0E+03
1.0E+02
1.0E+01
1.0E+00 1.0E-05
1.0E-03
1.0E-01
1.0E+01
1.0E+03
Reduced Frequency (Hz) Fig. 6. DM master curves at 21.1 °C.
1.0E+05
1.0E+07
649
J. Zhang et al. / Construction and Building Materials 111 (2016) 644–651
c0
1 dc AðKÞn
ð6Þ
where A and n are fracture properties determined from OT fracture test; c0 is initial crack length; h is asphalt layer thickness; K is stress intensity factor. Note the fatigue cracking model (Eq. (6)) incorporated in the TxME has been calibrated and validated using field performance data from the NCAT test track and numerous Texas field test sections [9]. Sensitivity analysis has shown that TxME can make rational predictions under different combinations of pavement structure, climate, and traffic loading [9]. Fatigue damage (D) is calculated using Miner’s law [9]:
D¼
X n Nf
ð7Þ
where n is actual load repetitions and Nf is the pavement allowable load repetitions. Then, fatigue cracking area is predicted using the equation below:
Fatigue cracking area ð%Þ ¼
100 1 þ e7:89 log D
ð8Þ
For the purpose of comparison, a typical pavement structure used in TxME is presented in Fig. 7. It consists of three layers: 101-mm HMA layer, 203-mm flexible base layer, and the subgrade. The thicknesses for three layers and moduli of flexible base and subgrade are all fixed while only the HMA material properties including DM and OT fracture properties obtained from the laboratory test were varied for the different mixes used in the HMA layer. A total of 3 million ESALs were used as traffic input for all the mixes and 50% was used as a failure criterion for fatigue cracking area [9]. Table 3 summarizes fracture properties (i.e., A and n) obtained from OT test and the predicted fatigue life corresponding to 50% fatigue cracked area. Fig. 8 presents the comparison of asphalt binder elastic recovery properties and the predicted fatigue life based on the OT fracture properties and DM data. As can be observed in the figure, D1, CA1, and F1 have the three greatest fatigue lives, which correspond to three highest asphalt binder elastic recovery values in ER test. When the asphalt binder ER elastic recovery value is smaller than 59%, there is no clear trend. It was also observed that CA1 and F1 with the same ER elastic recovery property (70%) have similar predicted fatigue lives. For the asphalt binder elastic recovery property measured in MSCR test, D1 has a low elastic recovery value while it exhibits the highest fatigue life. Due to the high testing temperature (64 °C) used in the MSCR test, it is not recommended to use it (MSCR) to correlate with fatigue performance based on OT fracture test performed at 25 °C.
Mix label
OT fracture properties A
n
1 2 3 4 5 6 7 8 9 10 11
D1 D2 D3 D4 D5 C1 C2 C3 B1 CA1 F1
1.316E07 5.830E06 3.932E04 1.330E04 1.330E04 4.204E06 1.913E06 3.067E06 5.199E04 1.447E06 1.838E07
4.735 3.898 3.272 3.968 3.968 3.875 4.130 3.566 3.302 4.812 4.594
Elastic Recovery_ER (%)
h
#
80 70 60 50 40 30 20 10 0
Fatigue life corresponding to 50% cracking area (months)
70
59%
59
D2
70
52
49
D1
169 99 56 110 86 110 132 83 57 158 157
23.5
23
23
D3
D4
D5
C1
29
26.8 27.1
C2
C3
B1 CA1 F1
(a) Elastic Recovery_MSCR0.1kPa (%)
Z Nf ¼
Table 3 OT fracture properties and predicted fatigue life corresponding to 50% cracked area.
80.0 70.0 60.0 50.0 40.0 30.0 20.0 10.0 0.0
71.8 61.2 42.2
42.0
32.9 16.0
D1
D2
D3
14.9 16.0
11.1 D4
41.4
37.5
D5
C1
C2
C3
B1 CA1
F1
(b) Predicted Fatigue Life (months)
typical highway pavement structure, TxME was utilized to predict fatigue cracking during the design period [9]. In the model used in TxME, crack propagation life Nf is computed based on the Paris’ law shown as follows [9]:
180 160 140 120 100 80 60 40 20 0
169
150 months
158 157
132 110
99
110 86
83 57
56
D1
D2
D3
D4
D5
C1
C2
C3
B1 CA1
F1
(c) Fig. 8. Comparisons of asphalt binder elastic recovery properties and TxME predicted fatigue life: (a) ER, (b) MSCR-0.1 kPa, (c) TxME predicted fatigue life.
Fig. 7. Hypothetical pavement structure used in TxME modeling.
The poor fatigue performance of the Type B mix, with only 57 months’ worth of service life, was expected. This is a coarsegraded mix, which is traditionally used as a base mix for rutting resistance purposes. It is rarely used for fatigue cracking resistance purposes in Texas, and historically known to be a poor fatigue crack-resistant mix which has been confirmed in the OT test (cycles to crack failure). For example, Walubita et al. (2010) conducted a study that for Type B mix, the OT cycles to failure is
650
J. Zhang et al. / Construction and Building Materials 111 (2016) 644–651
Fig. 9. Field performance after 48 months of service: (a) CAM, (b) Type F, (c) Type D, and (d) Type C.
smaller than 130 which fails the Texas OT pass-fail criterion of a minimum of 300 cycles [14]. In this study, it is used to validate the TxME fatigue performance prediction and also served as the ‘‘Control Mix”. On the other hand, the fine-graded Type F (F1) and CAM (CA1) mixes with the highest predicted pavement fatigue lives performed very well in the OT test (cycles to crack failure), which generally exhibit >900 cycles to failure [14]. In addition, the predicted fatigue lives for both CAM and Type F mixes are consistent with the observed satisfactory field performance. Fig. 9 shows no visual evidence of cracking after 48 months of service, which fairly correlates with the TxME fatigue predictions for CAM, Type F, Type D, and Type C mixes. 6. Summary and recommendations A laboratory study was undertaken to evaluate the relationship between the asphalt binder elastic recovery properties and HMA fatigue performance. The corresponding asphalt binder elastic recovery properties from the ER and MSCR tests were then correlated to the predicted HMA fatigue life based on the TxME modeling software, which was subsequently validated with other laboratory test and field performance. Two HMA performance tests (DM and OT fracture) were also conducted to generate the necessary HMA material properties for inputs into the TxME software. The key findings and recommendations are summarized as follows: There is no clear trend between the ER and MSCR test results for the measured asphalt binder elastic recovery property. The ER test is suggested to be used to correlate with HMA fatigue performance instead of the MSCR test due to inappropriate test temperature in the MSCR test. The ER test, on the other hand, is conducted at a relatively lower temperature where the HMA is more susceptible to cracking and poor fatigue performance.
HMA mixes with high asphalt binder elastic recovery properties generally exhibited low stiffness values. The TxME predicted fatigue performance life based on the OT fracture properties (A and n) was consistent with other laboratory test results and field observations. HMA mixes whose asphalt binders had high elastic recovery properties (P59%) showed long fatigue lives (>150 months), e.g. D1, CA1 and F1. However, no trends were observed for the other HMA mixes with low binder elastic recovery (<59%). To further evaluate the relationship between asphalt binder elastic recovery and HMA fatigue performance, it is recommended to investigate the effects of additives including different polymers, RAP, and RAS on the asphalt binder elastic recovery properties and their relationships with HMA fatigue performance in future studies. Acknowledgements The authors thank TxDOT for their financial support and all those who helped during the course of this research work. The contents of this paper reflect the views of the authors who were responsible for the facts and accuracy of the data presented herein and do not necessarily reflect the official views or policies of any agency or institute. This paper does not constitute a standard, specification, nor is it intended for design, construction, bidding, contracting, tendering, or permit purposes. References [1] A. Fatemi, L. Yang, Cumulative fatigue damage and life prediction theories: a survey of the state of the art for homogeneous materials, Int. J. Fatigue 20 (1) (1998) 9–34. [2] W. Mogawer, A. Austerman, M.E. Kutay, F. Zhou, Evaluation of binder elastic recovery on HMA fatigue cracking using continuum damage and overlay test based analyses, Road Mater. Pavement Des. 12 (2) (2011) 345–376.
J. Zhang et al. / Construction and Building Materials 111 (2016) 644–651 [3] AASHTO T301Elastic Recovery Test of Asphalt Materials by Means of a Ductilometer, American Association of State and Highway Transportation Officials, 2008. [4] AASHTO TP70-10Multiple Stress Creep Recovery (MSCR) Test of Asphalt Binder Using a Dynamic Shear Rheometer (DSR), American Association of State and Highway Transportation Officials, 2013. [5] Texas Department of Transportation, Online Manuals: The Overlay Test – Test Designation Tex-248-F, Austin, TX, USA, 2014 (accessed July 2015). [6] AASHTOStandard Specifications for Transportation Materials and Methods of Sampling and Testing. Standard TP 62-03. Standard Method of Test for Determining Dynamic Modulus of Hot Mix Asphalt Concrete Mixtures, Washington (DC), 2001. [7] T.K. Pellinen, M.W. Witczak, Stress dependent master curve construction for dynamic (complex) modulus, J. Assoc. Asphalt Paving Technol. 71 (2002). [8] S. Hu, F. Zhou, T. Scullion, J. Leidy, Calibrating and Validating Overlay TesterBased Fatigue Cracking Model with Data from National Center for Asphalt Technology, Transportation Research Record, Washington, DC, 2012, pp. 57–68.
651
[9] S. Hu, F. Zhou, T. Scullion, Development of Texas Mechanistic-Empirical Flexible Pavement Design System (TxME). 0-6622-2, Texas A&M Transportation Institute, College Station, TX, 2014. [10] P. Paris, F. Erdogan, A critical analysis of crack propagation laws, J. Basic Eng. 85 (1963) 528–534. [11] S. Hu, F. Zhou, T. Scullion, Implementation of Texas Asphalt Concrete Overlay Design System, Technical Report. 5-5123-03-1, Texas A&M Transportation Institute, College Station, TX, 2014. [12] Tex-242-FTxDOT-Texas Department of Transportation Designation. Hamburg Wheel Tracking Test. Austin, Texas, 2014 (http://ftp.dot.state.tx.us/pub/txdotinfo/cst/TMS/200-F_series/pdfs/bit242.pdf. Accessed on February 2015). [13] C.S. Clopotel, H.U. Bahia, Importance of elastic recovery in the dsr for binders and mastics, Eng. J. 16 (4) (2012) 99–106. [14] L.F. Walubita, G.S. Simate, H.O. Jeong, Characterizing the ductility and fatigue cracking resistance potential of asphalt mixes based on the laboratory direct tensile strength test, J. South African Inst. Civ. Eng. 52 (2) (2010) 31–40.
Contents lists available at ScienceDirect
Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat
Relating asphalt binder elastic recovery properties to HMA crack modeling and fatigue life prediction Jun Zhang a,⇑, Geoffrey S. Simate b, Sang Ick Lee c, Sheng Hu c, Lubinda F. Walubita c,d,e a
Department of Civil Engineering, Texas A&M University, College Station, TX 77843, USA University of the Witwatersrand, Johannesburg, P/Bag 3, Wits 2050, South Africa c Texas A&M Transportation Institute (TTI) – The Texas A&M University System, College Station, TX 77843, USA d Iliso (Z) Consulting Engineers Ltd, Lusaka, Zambia e Transportation Research Center, Wuhan Institute of Technology, Wuhan, China b
h i g h l i g h t s Binder elastic recovery is hypothesized to play a significant role in the fatigue cracking resistance. Two standardized tests are used to characterize binder elastic recovery properties. Two laboratory tests are used to characterize HMA properties of stiffness and fracture. HMA with high binder elastic recovery properties (P59%) has a long M-E predicted fatigue life. No clear trend is observed in predicted fatigue life for HMA with low elastic recovery (<59%).
a r t i c l e
i n f o
Article history: Received 11 November 2015 Received in revised form 4 February 2016 Accepted 23 February 2016 Available online 21 March 2016 Keywords: Asphalt binder Elastic recovery MSCR HMA Cracking Fracture properties Fatigue life Mechanistic-empirical (M-E) TxME
a b s t r a c t Fatigue cracking is one of the major distresses occurring in hot-mix asphalt (HMA) pavements, which is a consequence of accumulation of damage under repeated load applications and changes in the environmental conditions. HMA is predominantly composed of aggregates and asphalt binder, where the latter plays a significant role in the HMA fatigue performance. The elastic recovery of asphalt binders is one of the characteristic properties that are hypothesized to play a significant role in the fatigue cracking resistance of HMA pavements. In this study, the relationships between the asphalt binder elastic recovery properties and the HMA fatigue performance were comparatively investigated. Two asphalt binder tests, namely the elastic recovery (ER) and multiple stress creep recovery (MSCR) tests were conducted to comparatively characterize the asphalt binder elastic recovery properties in the laboratory, and their results were then correlated to the predicted HMA fatigue life based on the TxME modeling software, which was subsequently validated with other laboratory test results and field performance data. Eleven typical Texas HMA mixes collected from different construction sites were used for the study. Overall, the results indicated that HMA mixes with high asphalt binder elastic recovery properties (P59%) exhibited better cracking resistance potential with long predicted fatigue life (>150 months), which was also consistent with other studies and field performance observations. For mixes with low elastic recovery properties (<59%), there is no clear correlation with the predicted fatigue life. Ó 2016 Elsevier Ltd. All rights reserved.
1. Introduction Fatigue cracking is one of the major distresses occurring in hotmix asphalt (HMA) pavements, which is a consequence of accumulation of damage under repeated load applications [1]. HMA is predominantly composed of aggregates and asphalt binder, where the latter plays a significant role in the HMA fatigue performance. ⇑ Corresponding author. E-mail address: [email protected] (J. Zhang). http://dx.doi.org/10.1016/j.conbuildmat.2016.02.175 0950-0618/Ó 2016 Elsevier Ltd. All rights reserved.
Superpave Performance Grade (PG) specification addresses the asphalt binder property related to HMA fatigue performance by measuring a parameter G⁄sin d at intermediate temperatures, where G⁄ and d are the complex shear modulus and phase angle, respectively. However, this parameter was initially and primarily developed for non-modified asphalt binders [2]. With the current wide spread usage of modified asphalt binders, it is now becoming a challenge to effectively relate and quantify the expected HMA mix fatigue performance using the G⁄sin d parameter. Thus, this study was undertaken to investigate alternative asphalt binder
645
J. Zhang et al. / Construction and Building Materials 111 (2016) 644–651
properties relative to the HMA fatigue performance, namely the asphalt binder elastic recovery property. Polymer modified asphalt binders are widely used in HMA pavements, and one of the major improvements in these modified asphalt binders is the elastic recovery property, which is the degree to which an asphalt binder recovers to its original shape after release of the loading application. It is hypothesized that greater propensity to elastic recovery is desirable in HMA pavement to increase the fatigue cracking resistance performance [2]. Few studies have been conducted to investigate the relationship between the asphalt binder elastic recovery property and HMA fatigue characteristics [2]. Based on the aforementioned discussions, this study was initiated to investigate the relationship between the asphalt binder elastic recovery properties and HMA fatigue performance. To achieve this objective, two asphalt binder tests, namely elastic recovery (ER) and multiple stress creep recovery (MSCR); and two HMA performance tests, namely the, Dynamic Modulus (DM) and OT fracture test, were conducted [3–6]. The corresponding laboratory test results were then used to predict HMA fatigue life based on mechanistic-empirical (M-E) modeling with the Texas Mechanistic-Empirical (TxME) Flexible Pavement System software, which was subsequently validated with field performance data. Overall, this study was initiated to address the following three questions: (1) What is the relationship between ER and MSCR results for measured asphalt binder elastic recovery property? (2) What is the relationship between asphalt binder elastic recovery property and asphalt mixture stiffness? (3) What is the relationship between asphalt binder elastic recovery property and HMA fatigue performance? 2. Experimental plan Fig. 1 presents a schematic of the laboratory tests and the experimental research plan that was employed in this study. Two standardized asphalt binder tests, namely the ER and MSCR tests and two HMA performance tests including the DM and OT fracture were conducted. As illustrated in Fig. 1, the DM and OT fracture tests also provided input data for the M-E modeling with the TxME software in addition to characterizing the fundamental HMA properties such as modulus and fracture properties.
3. Laboratory test methods This section presents details of the laboratory tests employed in this study and includes the ER, MSCR, DM and OT fracture tests. 3.1. The ER test The elastic recovery test was performed in accordance with AASHTO T301 using a ductilometer and briquette specimens shown in Fig. 2 [3]. The specimens were conditioned in a water bath at 25 °C for one hour. The test was displacement-controlled and the asphalt binder specimens were pulled apart at a constant speed of 5 cm/min to an elongation of 20 cm, and held for 5 min. Then, a cut was made in the middle of the specimen and allowed them to remain in the ductilometer for recovery. After one hour, the percent elongation recovery was determined using the following equation:
% elongation ¼
20 x 100 20
ð1Þ
where x = final reading in centimeters after bringing the two severed ends of the specimen back together [3].
Fig. 1. Schematic illustration of the laboratory tests and experimental plan.
3.2. The MSCR test The MSCR test is creep and recovery test performed on an asphalt binder sample using the Dynamic Shear Rheometer (DSR) in accordance with AASHTO TP 70-10 [4]. The asphalt binder sample was 25 mm in diameter and 1 mm in thickness. The test was conducted at the high working temperature of the asphalt binder, which was controlled by the use of a water bath in the DSR machine setup. For each cycle, a Haversine shear load was applied to the sample for 1 s, followed by a 9-second rest period. Two stress levels of 0.1 kPa and 3.2 kPa were applied successively, and a total of 10 cycles were conducted for each stress level. A typical MSCR test results with two applied stress levels are shown in Fig. 3. As observed from the figure, the MSCR test characterizes the recovery properties of the asphalt binder under the shear creep load, which is the percent recovery. 3.3. The DM test Unconfined DM testing is an AASHTO standardized test method used for characterizing the stiffness, measured in terms of the dynamic complex modulus (|E⁄|), and visco-elastic properties of HMA mixes (AASHTO TP62-03) [6]. The DM is a stress-controlled test involving application of a repetitive sinusoidal dynamic compressive axial load (stress) to an unconfined HMA specimen over a range of different temperatures (i.e., 10, 4.4, 21.1, 37.8 and 54.4 °C) and loading frequencies (i.e., 0.1, 0.5, 1, 5, 10, 25 Hz) for each temperature. The typical parameter that results from the DM test is the |E⁄|, which was computed as:
jE j ¼
r0 e0
ð2Þ
where r0 is the compressive axial stress and e0 is the corresponding compressive axial resilient strain. For graphical analysis and easy interpretation of the DM test data, |E⁄| master-curves were also generated as a function of the loading frequency using the Pellinen et al.’s [7] time-temperature superposition sigmoidal model shown in Eqs. (3) and (4):
log jE j ¼ d þ
a
ð3Þ
1 þ ebc logðnÞ
logðnÞ ¼ logðf Þ þ logðaT Þ
ð4Þ ⁄
where n is the reduced frequency (Hz), d is the minimum |E | value (MPa), a is the span of |E⁄| values, and b and c are shape parameters. Parameters f and aT are the loading frequency and temperature shift
646
J. Zhang et al. / Construction and Building Materials 111 (2016) 644–651
Fig. 2. Elastic recovery test: (a) briquette specimens; (b) after elongation; (c) cut in the middle of the specimens after elongation.
(0.025 in.) and a loading rate of 10 s per cycle, composed of 5 s loading and 5 s unloading [5]. The test temperature was 25 °C [5]. The ‘‘OT Fracture” test procedure and setup is similar to the standard Tex-248-F illustrated in Fig. 4 except for some minor changes including the reduced opening displacement to 0.425 mm (0.017 in.) from 0.625 mm (0.025 in.) and that the test is run for a maximum of 100 cycles only. However, if the specimen fails in <50 cycles, then the opening displacement should be further reduced to 0.375 mm (0.015 in.) or less until it reaches a minimum of 50 cycles using different specimen sets [8]. The typical output that results from the OT Fracture test are the HMA fracture parameter A and n, which are used as inputs into the TxME software for fatigue life prediction [9]. At a fundamental level, these two parameters, A and n, compositely define the fracture behavior of HMA based on Paris’ law of fracture mechanics as illustrated in Eq. (5) [10].
dc ¼ AðDKÞn dN
Fig. 3. MSCR test: (a) asphalt binder sample loading configuration in the DSR; (b) 10 cycles of creep and recovery at two stress levels of 0.1 kPa and 3.2 kPa.
factor to temperature T, respectively. In this study, 21.1 °C was used as the reference temperature. 3.4. The OT fracture test The OT test is a simple performance test used to characterize the reflection cracking potential of HMA mixes in the laboratory in accordance with Texas specification Tex-248-F [5]. In this test, the HMA specimen was glued onto two metallic plates-one mobile and the other fixed during the testing process. As shown in Fig. 4, there is a 2-mm gap between the two plates to simulate the cracking on an existing old pavement underneath the new constructed pavement. The OT operates in a displacement-controlled loading mode to induce the horizontal movement in the mobile plate to simulate the opening and closing of cracking in the old pavements beneath an HMA overlay. Fig. 4 illustrates the standard Tex-248-F loading configuration consisting of a cyclic triangular displacementcontrolled waveform at a horizontal displacement of 0.625 mm
ð5Þ
where c is the crack length; N is the number of load cycles; dc/dN is the crack speed or rate of crack growth; DK is the change of stress intensity factor (SIF); and A and n are the fracture properties of the material, which are essentially regression coefficients in Eq. (5). Based on Eq. (5), it can be seen that the information required for determining the HMA fracture properties (A and n) are: 1) the crack length c corresponding to a specific number of load cycles N, and 2) the SIF (or DK) corresponding to a specific crack length c. In this study, a software called ‘‘OTA&n” that was developed for fitting the test data onto Paris’ Law model (Eq. (5)) to obtain the regression coefficients of A and n [11]. 4. Materials As listed in Table 1, a total of 11 Texas HMA mixes categorized in five mix types, namely Type B, C, D, F, and CAM, were collected from various field construction sites across Texas and used in this study. Asphalt binders were extracted from these plant-mixed materials and used for ER and MSCR testing. The plant-mix materials were compacted to an air void (AV) 7 ± 1% for HMA performance testing, namely DM and OT fracture [12]. All laboratory testing were conducted on a minimum of three replicate specimens. 5. Results and discussions In this section, the test results for the asphalt binders and HMA mixes are presented, analyzed, and discussed. The relationships between the asphalt binder elastic recovery and HMA fatigue related performance tests were investigated through quantitative
647
J. Zhang et al. / Construction and Building Materials 111 (2016) 644–651
Fig. 4. (a) The Overlay Tester (OT) and HMA specimen Setup; (b) OT loading configuration; (c) example of cracked HMA specimens after OT testing.
Table 1 HMA mix design characteristics. #
Mix label
Mix type
Aggregate gradation
Optimum asphalt content (%)
SHRP continuous grade
Total air voids (%)
1 2 3 4 5 6 7 8 9 10 11
D1 D2 D3 D4 D5 C1 C2 C3 B1 CA1 F1
Type Type Type Type Type Type Type Type Type CAM Type
Dense-graded (12.5 mm NMAS) Dense-graded (12.5 mm NMAS) Dense-graded (12.5 mm NMAS) Dense-graded (12.5 mm NMAS) Dense-graded (12.5 mm NMAS) Dense-graded (16 mm NMAS) Dense-graded (16 mm NMAS) Dense-graded (16 mm NMAS) Coarse-graded (22.4 mm NMAS) Fine-graded (9.5 mm NMAS) Fine-graded (9.5 mm NMAS)
5 5 5 5 5 5 5 5 5 7 6.7
76-29 81-28 70-32 70-32 86-31 78-27 74-30 74-31 83-31 82-34 76-29
7±1 7±1 7±1 7±1 7±1 7±1 7±1 7±1 7±1 7±1 7±1
D D D D D C C C B F
and graphical comparisons. Additionally, comparisons between the HMA fatigue related performance tests and the two asphalt binder recovery are also discussed. 5.1. Relationship between the ER and MSCR tests Table 2 presents the asphalt binder elastic recovery properties measured from the ER and MSCR tests. As noted in Table 2, the ER test was performed at a much lower temperature (25 °C) than MSCR test at 64 °C [3,4].
Fig. 5 shows the correlations of the ER and MSCR test for the asphalt binder elastic recovery. In a previous study conducted by Clopotel et al. (2012), a clear linear relationship was found between the elastic recovery measurements obtained from the ER and MSCR tests [13]. In Fig. 5(a) and (b), it shows a linear trend between the ER and MSCR, but the coefficient of correlation is relatively low (R2 6 0.60), especially for ER versus MSCR-3.2 kPa. Though both ER and MSCR tests measure the asphalt binder elastic recovery property, they are performed at different temperatures that may cause the poor correlation. It can be seen that there exists
648
J. Zhang et al. / Construction and Building Materials 111 (2016) 644–651
Table 2 ER and MSCR test results. Mix label
1 2 3 4 5 6 7 8 9 10 11
D1 D2 D3 D4 D5 C1 C2 C3 B1 CA1 F1
Elastic recovery property (%)
80.0 70.0 60.0 50.0 40.0 30.0 20.0 10.0 0.0
ER (10 °C)
MSCR-0.1 kPa (64 °C)
MSCR-3.2 kPa (64 °C)
59.0 49.0 23.5 23.0 23.0 52.0 29.0 26.8 27.1 70.0 70.0
32.9 42.2 16.0 11.1 42.0 37.5 14.9 16.0 41.4 71.8 61.2
14.9 37.0 4.1 1.2 39.6 24.6 7.4 4.0 42.5 63.4 48.4
y = 0.7539x + 14.591 R² = 0.6002
0.0
20.0
40.0
60.0
80.0
Elastic Recovery_MSCR-0.1kPa (%)
Elastic Recovery_ER (%)
(a) 80.0
5.2. Comparison of asphalt binder elastic recovery property and DM test results
y = 0.5518x + 26.723 R² = 0.3742
70.0 60.0 50.0 40.0 30.0 20.0 10.0 0.0
0.0
20.0
40.0
60.0
80.0
Elastic Recovery_MSCR-3.2kPa (%)
Elastic Recovery_MSCR3.2kPa (%)
(b) 70.0 y = 1.0445x - 10.66 R² = 0.9376
60.0 50.0
Fig. 6 presents the DM (|E⁄|) master curves derived using Eqs. (3) and (4). As can be seen in the figure, CA1 and F1 with the greatest asphalt binder elastic recovery values exhibit lower stiffness when compared to the other mixes. Similarly, D1 has the third largest ER elastic recovery value, and it shows a lower stiffness compared to most of the other mixes. It should be noted that the lower stiffness of CA1 and F1 may be partially attributed to higher asphalt contents as well as higher binder elastic recovery. In addition, DM test results were used as inputs for pavement fatigue performance prediction based on the OT fracture properties in the following section. 5.3. Comparison of the asphalt binder elastic recovery property and predicted pavement fatigue performance based on OT fracture properties
40.0 30.0
In order to evaluate the correlation between the asphalt binder elastic recovery property and pavement fatigue performance of a
20.0 10.0 0.0
0.0
20.0
40.0
60.0
80.0
Elastic Recovery_MSCR-0.1kPa (%)
(c) Fig. 5. Correlations: (a) ER versus MSCR-0.1 kPa; (b) ER versus MSCR-3.2 kPa; (c) MSCR-0.1 versus MSCR-3.2 kPa.
a good linear correlation between the MSCR-0.1 kPa and MSCR3.2 kPa, which may be attributed to the same testing temperature. This also means that for the same temperature, the asphalt binder elastic recovery property at either of two stress levels of the MSCR test can be estimated if the other is known as they have a clear linear relationship (R2 = 0.93).
1.0E+04
Dynamic Modulus (Mpa)
Elastic Recovery_ER (%)
#
Theoretically, the linear correlation exhibited in Fig. 5(c) was expected. This is because, other than the differences in the stress levels (i.e., magnitude), the test loading mode/configuration and temperature were the same since both are MSCR tests. Therefore, the response behavior and ER trend of the different asphalt binders under the two different stress levels (0.1 and 3.2 kPa) of the MSCR test were theoretically expected to be similar as exemplified in Fig. 5(c), with a clear linear relationship and an R2 value of 93%. Other than from the literature data by Clopotel et al. (2012), the relationship between the ER and MSCR tests was not really anticipated and theoretically unknown. For the tested asphalt binders in Fig. 5(a) and (b), the linearity in the correlation is not very strong with the MSCR-0.1 kPa (R2 = 0.60) out-performing the MSCR3.2 kPa (R2 = 0.37). Thus, the MSCR-0.1 kPa would be preferred over the MSCR-3.2 kPa in terms of correlation with the ER test. As previously stated, this poor correlation could be attributed to the differences in the test loading modes (configurations) and test temperatures (i.e., ER at 10 °C and MSCR at 64 °C). By and large, however, the particular asphalt binders (i.e., types, grades, etc.) and the number of asphalt binders or sample replicates (i.e., number of data points) selected for this study could have had an influence on the graphical correlations in Fig. 5(a) and (b). Although it has been concluded from Fig. 5 that the MSCR-0.1 kPa (R2 = 0.60) is superior to the MSCR-3.2 kPa (R2 = 0.37) in terms of correlating with the ER test data, the challenge with the data limitation and relatively poor R2 values suggest additional testing with a wide array of asphalt binders as well as exploring other correlative criteria for the ER-MSCR relationships in future studies.
D1 D2 D3 D4 D5 C1 C2 C3 B1 CA1 F1
1.0E+03
1.0E+02
1.0E+01
1.0E+00 1.0E-05
1.0E-03
1.0E-01
1.0E+01
1.0E+03
Reduced Frequency (Hz) Fig. 6. DM master curves at 21.1 °C.
1.0E+05
1.0E+07
649
J. Zhang et al. / Construction and Building Materials 111 (2016) 644–651
c0
1 dc AðKÞn
ð6Þ
where A and n are fracture properties determined from OT fracture test; c0 is initial crack length; h is asphalt layer thickness; K is stress intensity factor. Note the fatigue cracking model (Eq. (6)) incorporated in the TxME has been calibrated and validated using field performance data from the NCAT test track and numerous Texas field test sections [9]. Sensitivity analysis has shown that TxME can make rational predictions under different combinations of pavement structure, climate, and traffic loading [9]. Fatigue damage (D) is calculated using Miner’s law [9]:
D¼
X n Nf
ð7Þ
where n is actual load repetitions and Nf is the pavement allowable load repetitions. Then, fatigue cracking area is predicted using the equation below:
Fatigue cracking area ð%Þ ¼
100 1 þ e7:89 log D
ð8Þ
For the purpose of comparison, a typical pavement structure used in TxME is presented in Fig. 7. It consists of three layers: 101-mm HMA layer, 203-mm flexible base layer, and the subgrade. The thicknesses for three layers and moduli of flexible base and subgrade are all fixed while only the HMA material properties including DM and OT fracture properties obtained from the laboratory test were varied for the different mixes used in the HMA layer. A total of 3 million ESALs were used as traffic input for all the mixes and 50% was used as a failure criterion for fatigue cracking area [9]. Table 3 summarizes fracture properties (i.e., A and n) obtained from OT test and the predicted fatigue life corresponding to 50% fatigue cracked area. Fig. 8 presents the comparison of asphalt binder elastic recovery properties and the predicted fatigue life based on the OT fracture properties and DM data. As can be observed in the figure, D1, CA1, and F1 have the three greatest fatigue lives, which correspond to three highest asphalt binder elastic recovery values in ER test. When the asphalt binder ER elastic recovery value is smaller than 59%, there is no clear trend. It was also observed that CA1 and F1 with the same ER elastic recovery property (70%) have similar predicted fatigue lives. For the asphalt binder elastic recovery property measured in MSCR test, D1 has a low elastic recovery value while it exhibits the highest fatigue life. Due to the high testing temperature (64 °C) used in the MSCR test, it is not recommended to use it (MSCR) to correlate with fatigue performance based on OT fracture test performed at 25 °C.
Mix label
OT fracture properties A
n
1 2 3 4 5 6 7 8 9 10 11
D1 D2 D3 D4 D5 C1 C2 C3 B1 CA1 F1
1.316E07 5.830E06 3.932E04 1.330E04 1.330E04 4.204E06 1.913E06 3.067E06 5.199E04 1.447E06 1.838E07
4.735 3.898 3.272 3.968 3.968 3.875 4.130 3.566 3.302 4.812 4.594
Elastic Recovery_ER (%)
h
#
80 70 60 50 40 30 20 10 0
Fatigue life corresponding to 50% cracking area (months)
70
59%
59
D2
70
52
49
D1
169 99 56 110 86 110 132 83 57 158 157
23.5
23
23
D3
D4
D5
C1
29
26.8 27.1
C2
C3
B1 CA1 F1
(a) Elastic Recovery_MSCR0.1kPa (%)
Z Nf ¼
Table 3 OT fracture properties and predicted fatigue life corresponding to 50% cracked area.
80.0 70.0 60.0 50.0 40.0 30.0 20.0 10.0 0.0
71.8 61.2 42.2
42.0
32.9 16.0
D1
D2
D3
14.9 16.0
11.1 D4
41.4
37.5
D5
C1
C2
C3
B1 CA1
F1
(b) Predicted Fatigue Life (months)
typical highway pavement structure, TxME was utilized to predict fatigue cracking during the design period [9]. In the model used in TxME, crack propagation life Nf is computed based on the Paris’ law shown as follows [9]:
180 160 140 120 100 80 60 40 20 0
169
150 months
158 157
132 110
99
110 86
83 57
56
D1
D2
D3
D4
D5
C1
C2
C3
B1 CA1
F1
(c) Fig. 8. Comparisons of asphalt binder elastic recovery properties and TxME predicted fatigue life: (a) ER, (b) MSCR-0.1 kPa, (c) TxME predicted fatigue life.
Fig. 7. Hypothetical pavement structure used in TxME modeling.
The poor fatigue performance of the Type B mix, with only 57 months’ worth of service life, was expected. This is a coarsegraded mix, which is traditionally used as a base mix for rutting resistance purposes. It is rarely used for fatigue cracking resistance purposes in Texas, and historically known to be a poor fatigue crack-resistant mix which has been confirmed in the OT test (cycles to crack failure). For example, Walubita et al. (2010) conducted a study that for Type B mix, the OT cycles to failure is
650
J. Zhang et al. / Construction and Building Materials 111 (2016) 644–651
Fig. 9. Field performance after 48 months of service: (a) CAM, (b) Type F, (c) Type D, and (d) Type C.
smaller than 130 which fails the Texas OT pass-fail criterion of a minimum of 300 cycles [14]. In this study, it is used to validate the TxME fatigue performance prediction and also served as the ‘‘Control Mix”. On the other hand, the fine-graded Type F (F1) and CAM (CA1) mixes with the highest predicted pavement fatigue lives performed very well in the OT test (cycles to crack failure), which generally exhibit >900 cycles to failure [14]. In addition, the predicted fatigue lives for both CAM and Type F mixes are consistent with the observed satisfactory field performance. Fig. 9 shows no visual evidence of cracking after 48 months of service, which fairly correlates with the TxME fatigue predictions for CAM, Type F, Type D, and Type C mixes. 6. Summary and recommendations A laboratory study was undertaken to evaluate the relationship between the asphalt binder elastic recovery properties and HMA fatigue performance. The corresponding asphalt binder elastic recovery properties from the ER and MSCR tests were then correlated to the predicted HMA fatigue life based on the TxME modeling software, which was subsequently validated with other laboratory test and field performance. Two HMA performance tests (DM and OT fracture) were also conducted to generate the necessary HMA material properties for inputs into the TxME software. The key findings and recommendations are summarized as follows: There is no clear trend between the ER and MSCR test results for the measured asphalt binder elastic recovery property. The ER test is suggested to be used to correlate with HMA fatigue performance instead of the MSCR test due to inappropriate test temperature in the MSCR test. The ER test, on the other hand, is conducted at a relatively lower temperature where the HMA is more susceptible to cracking and poor fatigue performance.
HMA mixes with high asphalt binder elastic recovery properties generally exhibited low stiffness values. The TxME predicted fatigue performance life based on the OT fracture properties (A and n) was consistent with other laboratory test results and field observations. HMA mixes whose asphalt binders had high elastic recovery properties (P59%) showed long fatigue lives (>150 months), e.g. D1, CA1 and F1. However, no trends were observed for the other HMA mixes with low binder elastic recovery (<59%). To further evaluate the relationship between asphalt binder elastic recovery and HMA fatigue performance, it is recommended to investigate the effects of additives including different polymers, RAP, and RAS on the asphalt binder elastic recovery properties and their relationships with HMA fatigue performance in future studies. Acknowledgements The authors thank TxDOT for their financial support and all those who helped during the course of this research work. The contents of this paper reflect the views of the authors who were responsible for the facts and accuracy of the data presented herein and do not necessarily reflect the official views or policies of any agency or institute. This paper does not constitute a standard, specification, nor is it intended for design, construction, bidding, contracting, tendering, or permit purposes. References [1] A. Fatemi, L. Yang, Cumulative fatigue damage and life prediction theories: a survey of the state of the art for homogeneous materials, Int. J. Fatigue 20 (1) (1998) 9–34. [2] W. Mogawer, A. Austerman, M.E. Kutay, F. Zhou, Evaluation of binder elastic recovery on HMA fatigue cracking using continuum damage and overlay test based analyses, Road Mater. Pavement Des. 12 (2) (2011) 345–376.
J. Zhang et al. / Construction and Building Materials 111 (2016) 644–651 [3] AASHTO T301Elastic Recovery Test of Asphalt Materials by Means of a Ductilometer, American Association of State and Highway Transportation Officials, 2008. [4] AASHTO TP70-10Multiple Stress Creep Recovery (MSCR) Test of Asphalt Binder Using a Dynamic Shear Rheometer (DSR), American Association of State and Highway Transportation Officials, 2013. [5] Texas Department of Transportation, Online Manuals: The Overlay Test – Test Designation Tex-248-F, Austin, TX, USA, 2014 (accessed July 2015). [6] AASHTOStandard Specifications for Transportation Materials and Methods of Sampling and Testing. Standard TP 62-03. Standard Method of Test for Determining Dynamic Modulus of Hot Mix Asphalt Concrete Mixtures, Washington (DC), 2001. [7] T.K. Pellinen, M.W. Witczak, Stress dependent master curve construction for dynamic (complex) modulus, J. Assoc. Asphalt Paving Technol. 71 (2002). [8] S. Hu, F. Zhou, T. Scullion, J. Leidy, Calibrating and Validating Overlay TesterBased Fatigue Cracking Model with Data from National Center for Asphalt Technology, Transportation Research Record, Washington, DC, 2012, pp. 57–68.
651
[9] S. Hu, F. Zhou, T. Scullion, Development of Texas Mechanistic-Empirical Flexible Pavement Design System (TxME). 0-6622-2, Texas A&M Transportation Institute, College Station, TX, 2014. [10] P. Paris, F. Erdogan, A critical analysis of crack propagation laws, J. Basic Eng. 85 (1963) 528–534. [11] S. Hu, F. Zhou, T. Scullion, Implementation of Texas Asphalt Concrete Overlay Design System, Technical Report. 5-5123-03-1, Texas A&M Transportation Institute, College Station, TX, 2014. [12] Tex-242-FTxDOT-Texas Department of Transportation Designation. Hamburg Wheel Tracking Test. Austin, Texas, 2014 (http://ftp.dot.state.tx.us/pub/txdotinfo/cst/TMS/200-F_series/pdfs/bit242.pdf. Accessed on February 2015). [13] C.S. Clopotel, H.U. Bahia, Importance of elastic recovery in the dsr for binders and mastics, Eng. J. 16 (4) (2012) 99–106. [14] L.F. Walubita, G.S. Simate, H.O. Jeong, Characterizing the ductility and fatigue cracking resistance potential of asphalt mixes based on the laboratory direct tensile strength test, J. South African Inst. Civ. Eng. 52 (2) (2010) 31–40.