Evaluating fatigue behavior of asphalt mixtures under alternate tension–compression loading model using new alternate biaxial splitting method Zhesheng Ge a , Hao Wang b , Yangyang Wang a , Xiaoqian Hu a a State Key Laboratory of Subtropical Building Science, South China University of Technology, Wushan, Tianhe, Guangzhou, Guangdong 510640, China b Rutgers, The State University of New Jersey, Piscataway, NJ 08854, United States highlights An alternative biaxial splitting (ABS) fixture has been developed. Fatigue behavior of asphalt mixtures under a tension–compression loading mode has been analyzed. The laboratory fatigue life of the ABS test samples was longer than that of the IDT test. The creep effect of compressive stress produced increased fatigue life of asphalt mixtures. graphical abstract Top plate Joint 100 135° Jack bolt Jack bolt Bottom loading strip Right loading strip Guide rod Left loading strip Connecting plate Top loading strip Bottom plate article info Article history: Received 4 October 2013 Received in revised form 6 December 2013 Accepted 16 December 2013 Keywords: Asphalt mixtures Fatigue behavior Test method Tension–compression Alternate biaxial splitting. abstract The primary purpose of this study is to develop a new test method with a self-developed alternate biaxial splitting (ABS) fixture for evaluating fatigue behavior of asphalt mixtures under a tension–compression loading mode using new alternate biaxial splitting method. A comparison was performed in the indirect tension (IDT) and ABS test on laboratory-mixed asphalt concrete (AC-13) with an optimum asphalt con- tent of 4.7%. The results showed that the strain growth rate of the asphalt samples in the ABS fatigue test increased slower than that of in the IDT fatigue test with an increase in loading cycles. When the damage evolution in the asphalt samples under fatigue loading reached the steady state, the ratio of damage change (RDC) of IDT fatigue testing samples changed little maintaining greater than zero all the time. In contrast, the RDC of ABS fatigue testing samples had an alternate positive–negative variation. The laboratory fatigue life of the ABS test was 4–23 times more than that of the IDT test at a constant tem- perature of 20 °C. It demonstrated that the creep effect of compressive stress promoted the healing of micro-cracks and produced increased fatigue life. Ó 2013 Elsevier Ltd. All rights reserved. 1. Introduction Fatigue cracking is one of the most influential distresses that govern the service life of asphalt pavements. Fatigue is a phenom- enon in which a pavement is subjected to cyclic stress levels typi- cally less than the ultimate failure stress. Overall understanding of this cracking phenomenon suggests that these cycles create areas of tensile stresses at the bottom of the pavement layer, which cause the initiation of micro-cracks. Under repeated loadings, these micro-cracks densify, propagate, and eventually develop into more visible macro-cracks on the pavement surface [1–3]. A precise 0950-0618/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.conbuildmat.2013.12.042 E-mail addresses: [email protected](Z. Ge), [email protected](H. Wang), [email protected](Y. Wang), [email protected](X. Hu) Construction and Building Materials 54 (2014) 106–112 Contents lists available at ScienceDirect Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat
Transcript
Construction and Building Materials 54 (2014) 106–112
Zhesheng Ge a, Hao Wang b, Yangyang Wang a, Xiaoqian Hu a
a State Key Laboratory of Subtropical Building Science, South China University of Technology, Wushan, Tianhe, Guangzhou, Guangdong 510640, Chinab Rutgers, The State University of New Jersey, Piscataway, NJ 08854, United States
h i g h l i g h t s
� An alternative biaxial splitting (ABS)fixture has been developed.� Fatigue behavior of asphalt mixtures
under a tension–compression loadingmode has been analyzed.� The laboratory fatigue life of the ABS
test samples was longer than that ofthe IDT test.� The creep effect of compressive stress
produced increased fatigue life ofasphalt mixtures.
g r a p h i c a l a b s t r a c t
Top plate
Joint
100
135°
Jack boltJack bolt
Bottom loading strip
Right loading strip
Guide rod
Left loading strip
Connecting plate
Top loading strip
Bottom plate
a r t i c l e i n f o
Article history:Received 4 October 2013Received in revised form 6 December 2013Accepted 16 December 2013
The primary purpose of this study is to develop a new test method with a self-developed alternate biaxialsplitting (ABS) fixture for evaluating fatigue behavior of asphalt mixtures under a tension–compressionloading mode using new alternate biaxial splitting method. A comparison was performed in the indirecttension (IDT) and ABS test on laboratory-mixed asphalt concrete (AC-13) with an optimum asphalt con-tent of 4.7%. The results showed that the strain growth rate of the asphalt samples in the ABS fatigue testincreased slower than that of in the IDT fatigue test with an increase in loading cycles. When the damageevolution in the asphalt samples under fatigue loading reached the steady state, the ratio of damagechange (RDC) of IDT fatigue testing samples changed little maintaining greater than zero all the time.In contrast, the RDC of ABS fatigue testing samples had an alternate positive–negative variation. Thelaboratory fatigue life of the ABS test was 4–23 times more than that of the IDT test at a constant tem-perature of 20 �C. It demonstrated that the creep effect of compressive stress promoted the healing ofmicro-cracks and produced increased fatigue life.
� 2013 Elsevier Ltd. All rights reserved.
1. Introduction
Fatigue cracking is one of the most influential distresses thatgovern the service life of asphalt pavements. Fatigue is a phenom-
enon in which a pavement is subjected to cyclic stress levels typi-cally less than the ultimate failure stress. Overall understanding ofthis cracking phenomenon suggests that these cycles create areasof tensile stresses at the bottom of the pavement layer, whichcause the initiation of micro-cracks. Under repeated loadings, thesemicro-cracks densify, propagate, and eventually develop into morevisible macro-cracks on the pavement surface [1–3]. A precise
where Gm is the bulk specific gravity; Gt the theoretical specific gravity; Vv the airvoids (%); VMA the voids in mineral aggregate (%); and VFA is the voids filled withasphalt (%).
Z. Ge et al. / Construction and Building Materials 54 (2014) 106–112 107
understanding of the fatigue behavior of asphalt mixtures is re-quired in order to improve asphalt mixture design and perfor-mance. However, accurate prediction and evaluation of fatigue isa difficult task not only because of the complex nature of fatiguephenomena but also because of characteristics of fatigue testing[4–6]. Over the past 40 years, many diverse fatigue tests have beendeveloped to simulate the fatigue behavior of asphalt mixtureswith varying success. These fatigue tests can be basically catego-rized into the following types according to experimental samples,
100
(a) Front view of the sample.
Fig. 1. Representational samples used i
Joint
100
135°
Jack bolt
Bottom loadin
Guide rod
Left loading strip
Fig. 2. Front view of the ABS fixture (uni
beam sample (two-, three-, and four-point bending and fracturemechanics tests), cylinder sample (direct tension, indirect tensionand triaxial tests), full-scale testing pavement (wheel trackingtest). Among the above fatigue tests, the indirect tension (IDT) testis a practical and cheap configuration for testing of asphalt con-crete, as asphalt concrete samples are often cylindrical in shape.Moreover, the cylinder sample can be produced by a laboratorycylindrical shaped mold during compaction or taken from the fieldby a core barrel. Considering from the cost, simplicity andconvenience, the IDT fatigue test had been adopted widely by
50
(b) Side view of the sample.
n the fatigue test (scale unit, mm).
Top plate
Jack bolt
g strip
Right loading strip
Connecting plate
Top loading strip
Bottom plate
ts: degree for angle, mm for length).
3520.8
0 0.05 0.1 0.15 0.2 Time/s
Loa
d/N
20
(a) Vertical loading waveform of the IDT fatigue test.
-3520.8
0
3520.8
Time/s
Loa
d/N
0 0.05 0.1 0.15 0.2
Vertical loadHorizontal load
(b) Loading waveform of the ABS fatigue test.
Fig. 3. Loading waveforms at 10 Hz with R = 0.3.
Fig. 4. Phase angle values for sample AC-13 at 20 �C.
108 Z. Ge et al. / Construction and Building Materials 54 (2014) 106–112
researchers [7]. But the load applied by the vehicle is in nature acontinuously changing moving load as the vehicle is approachingand leaving, the longitudinal strain at the bottom of the asphaltlayer is composed of a compressive part followed by a tensile partand another compressive part. However, the IDT fatigue test can-not provide an alternate tension–compression loading mode, so fa-tigue life from laboratory experimental data under-predicted fieldobservations [8]. This difference has typically been accounted forby using a single shift factor based on engineering experience. Fourshift factors can be identified: stress state, traffic wander, asphaltmixtures healing and material properties [9].
Although many studies have devoted to fatigue test of asphaltmixtures, very few researchers have developed a kind of fixturewhich is simple and consistent to the response of pavement andproduced a more accurate loading conditions for laboratory fatiguetest. Due to the tension–compression stress state of the pavementat varying temperatures and the moving load pattern (approachingand leaving), accordingly, the concept has been applied here to putforward a new fatigue evaluation test.
The objectives of this paper are:
(a) To develop an alternate biaxial splitting (ABS) method toevaluate fatigue behavior of asphalt mixtures under a ten-sion–compression loading mode.
(b) To compare results obtained from the proposed ABS testwith results from the IDT fatigue test.
2. Materials
The asphalt mixture tested in this study is AC-13 with an optimum asphalt con-tent of 4.7%. The AC-13 asphalt mixture was first molded to cylindrical samples of150 mm in diameter in 100 gyrations using the Superpave Gyratory Compactor(SGC) then cored to obtain the 100 mm diameter cylindrical testing samples outof the middle section of the SGC samples that were individually cut to a thicknessof 50 mm. The volume index of AC-13 is shown in Table 1, and the representationalsamples used in the fatigue test are shown in Fig. 1.
3. Test methods
3.1. The ABS fixture
The alternate biaxial splitting (ABS) fixture used in this study isshown in Fig. 2. The regular octagonal ABS fixture has a vertical ac-tive loading part and a horizontal passive loading part. The twoloading parts are connected by jointed connecting plates. The topplate is the vertical active loading part which is connected to theload cell on the crosshead and must be placed horizontally to keepit running along the guide rods vertically and smoothly. The innercurve diameter of the ABS fixture is 100 mm and the included an-gle was initialized to 135�, so the horizontal loading pressure isequal to the applied vertical tension. The top, bottom, and left load-ing strips are fixed but the right loading strip is adjustable. Thewidth of the loading strips was 12.7 mm. To keep the sample lo-cated in the middle of the ABS fixture, the length of the loadingstrips is 20 mm longer than the height of the sample. The edgesof the loading strips were rounded out by mechanical grinding toavoid snagging the sample.
The installation procedure of the cylinder sample involves: (1)loosening the right loading strip, (2) inserting the cylinder sampleinto the center of the ABS fixture, (3) setting the distance betweenthe top and the bottom strips to 100 mm by adjusting the jackbolts, (4) fastening the right loading strip until the distance be-tween the left and right reaches 100 mm, and lastly (5) adjustingthe jack bolts downward at least 10 mm.
When a vertical pressure is applied on the top plate of the ABSfixture, the vertical loading strips will transfer the pressure to thesample, meanwhile the horizontal loading strips will separate fromthe sample, so the sample suffers the vertical splitting; when a ver-tical tension is applied on the top plate of the ABS fixture, the ver-tical loading strips will separate from the sample, meanwhile thehorizontal loading strips will transfer the pressure to the sample,so the sample suffers the horizontal splitting. By repeating thiskind of loading cycle, the sample will suffer repeated, alternatetension–compression stress.
3.2. Experimental parameters
Both the IDT tests and the ABS fatigue tests were performedusing the material testing systems (MTS) servo-hydraulic closed-loop testing machine. An environmental chamber was used tomaintain the temperature of the samples at 20 �C. Prior to testing,the sample was placed into the chamber at the testing temperaturefor at least two hours for conditioning. The tests were performedunder stress-controlled conditions according to four stress ratios(namely, R = 0.3, R = 0.4, R = 0.5 and R = 0.6) which means the ratioof given to maximum stress in fatigue testing. A sinusoidal wave-form load was applied in the fatigue tests at frequencies of 5 Hzand 10 Hz. Failure of a sample tested in the fatigue test can be de-fined at the time when the maximum vertical deformation reachedto 5 mm.
Prior to the fatigue tests, the splitting strength of the AC-13samples was tested to get the loading value that was applied tothe fatigue samples under different stress ratios. The splitting test
(a) Failure mode in the IDT fatigue test.
(b) Failure mode in the ABS fatigue test.
Fig. 5. Failure modes in the IDT and ABS fatigue tests.
Z. Ge et al. / Construction and Building Materials 54 (2014) 106–112 109
was conducted at 20 �C with a constant loading rate of 50 mm/min.The results of the spitting strength are shown in Table 2.
The loading waveforms at 10 Hz with R = 0.3 for the samples inthe IDT and ABS fatigue tests are shown in Fig. 3. In the ABS fatigue
tests, the continuous sinusoidal loading waveform was applied inthe top plate first then the load was transferred to the horizontalloading strip, thus the loading waveform for each sample was com-posed of a vertical compressive part followed by a horizontal
ε
Fig. 6. R = 0.4, e–N curves.
Fig. 7. R = 0.6, e–N curves in the ABS fatigue test.
Table 3Mathematical functions of e–N curves.
Stressratio
Type of fatiguetest
Functions Correlationcoefficient
0.4 IDT e = 0.0056 exp (0.0043N) 0.96710.4 ABS e = 0.0022 exp (0.0009N) 0.99060.6 IDT e = 0.0046 exp (0.0136N) 0.99120.6 ABS e = 0.0035 exp (0.0022N) 0.8711
110 Z. Ge et al. / Construction and Building Materials 54 (2014) 106–112
compressive part and another vertical compressive part such thatthe phase difference between the vertical and horizontal loadingwaveform was half cycle.
4. Results and discussion
4.1. Phase angle of ABS fatigue testing samples
The phase angle, /, is related to the time lag, Dt, between thestress input and the strain response.
/ ¼ 2pfDt ð1Þ
where f is the loading frequency, Hz.For f = 5 Hz and R = 0.6, phase angle values for the sample AC-13
at 20 �C are shown in Fig. 4.Seen from Fig. 4, the phase angle u distributed loosely and had
no regulations to clarify its characteristics at the initial period. Be-cause the ABS fatigue sample was in an unsteady state when a loadwas suddenly applied on it and the inner structure of the ABS fati-gue sample was rearranging at the initial loading period. After cer-tain cycles of loading, the phase angle basically had an increasingtrend. It meant that the asphalt samples tended to be more viscous,moreover, the more viscous the asphalt sample is, and the moreenergy it would dissipate and it becomes easier to fracture.
4.2. Comparison between the IDT and ABS fatigue tests
4.2.1. Fracture characteristicsAs shown in Fig. 5, both fatigue failure modes for the IDT and
ABS fatigue tests were investigated.From Fig. 5, it shows that the cylinder asphalt samples have a
fixed failure mode in the IDT fatigue test which is cracking alongthe spitting direction but the failure mode in the ABS fatigue testis various and random—such as vertical cracking, horizontal crack-ing, or both vertical and horizontal cracking—bearing a closerresemblance to pavement micro-cracking observed in the field.This is because at both vertical and horizontal planes the compres-sion–tension cycles are applied alternately.
Fatig
ue li
fe (
cycl
e)
Stress ratio
The ABS fatigue testThe IDT fatigue test
(a) Fatigue life from the IDT and ABS fatigue tests (10Hz).
Fatig
ue li
fe (
cycl
e)
Stress ratio
The ABS fatigue testThe IDT fatigue test
(b) Fatigue life from the IDT and ABS fatigue tests (5Hz).
Fig. 9. Fatigue life from the IDT and ABS fatigue test.
Fatig
ue li
fe r
atio
Stress ratio
Fig. 10. Fatigue life ratios between the ABS and IDT fatigue test.
Z. Ge et al. / Construction and Building Materials 54 (2014) 106–112 111
4.2.2. Analysis of vertical strainVertical strains were measured by the vertical displacement
sensor of the MTS. The vertical strain and the number of load cycles
were named e and N respectively. The e–N curves obtained for thecylinder samples at 20 �C and 5 Hz under constant stress ampli-tude for R = 0.4 and R = 0.6 are plotted in Figs. 6 and 7.
The e–N curves in both fatigue tests could be divided into threedifferent stages—the initial stage, the steady stage, and the ascentstage. In the initial stage, e in the ABS fatigue test changed littlewhereas e in the IDT fatigue test maintained a state of lineargrowth. In the steady stage, e in the ABS fatigue test grew slowlywhereas e in the IDT fatigue test continued a state of linear growth.In the ascent stage, however, the e in both fatigue tests grew fastuntil the cylinder samples fractured.
Mathematical functions of e–N curves were shown in Table 3.The e–N curves of the IDT fatigue test grew faster than that ofthe ABS fatigue test because the cylinder sample was not in analternate tension–compression stress state like the ABS fatiguetest. The horizontal compressive stress possibly contributed tothe healing of the fatigue micro-cracking.
4.2.3. Analysis of the ratio of damage changeMany studies have devoted to fatigue damage of asphalt
mixtures [10]. In this study, the damage factor, namely D, wasdefined as,
D ¼ 1� SN
S0ð2Þ
Using D as the evaluation parameter and examining the ratio ofdamage change (RDC) with respect to each loading cycle in thefatigue test, this approach is similar to the ratio of dissipatedenergy change (RDEC) approach [10]. Therefore, the RDC betweenload cycles n and n + 1 can be represented as,
RDCi ¼Diþn � Di
nð3Þ
where D is the damage factor; S0 the initial stiffness modulus(MPa); SN the stiffness modulus produced in load cycle n (MPa);RDCi the ratio of damage change from load cycle n to load cyclen + 1; Di the damage factor produced in load cycle n; and Di+n isthe damage factor produced in load cycle n + 1.
Fig. 8. shows RDCi versus load cycle (n) for the samples at 10 Hzand a stress ratio of R = 0.6.
Fig. 8(a) has three different stages. The RDCi decreased rapidlyin the first stage. During the second stage, the RDCi came into asteady state and changed very little during a long period of time.Finally, in the beginning of the third stage, the RDCi increased rap-idly. Note: the RDCi for all three stages was always greater thanzero. Fig. 8(b) can be divided into four stages. The RDCi decreasedrapidly in the first stage. During the second stage, the RDCi changedsteadily alternating between positive values and negative valuesduring a long period of time. In the third stage, the RDCi stayed po-sitive. Finally, in the beginning of the fourth stage, the RDCi in-creased rapidly. The second stage of this ABS fatigue testdemonstrated that the horizontal compressive stress can indeedcontribute to the healing of the asphalt samples.
4.2.4. Analysis of fatigue lifeThe fatigue life of AC-13samples tested by the IDT and the ABS
fatigue tests is shown in Fig. 9. The ratios of fatigue life betweenthe ABS and the IDT fatigue tests are shown in Fig. 10 (arrangedaccording to the different stress ratios and frequencies).
Fig. 9 indicated that fatigue life of the AC-13 sample from theABS fatigue test was longer than that from the IDT fatigue test.Fig. 10 indicates that the fatigue life from the ABS fatigue test is4–12 times and 12–23 times more than that from the IDT test at5 Hz and 10 Hz respectively.
Recent studies indicated that the fatigue life from an IDT testwas shorter than the field fatigue life of asphalt concrete because
112 Z. Ge et al. / Construction and Building Materials 54 (2014) 106–112
the longitudinal strain at the bottom of the asphalt layer was com-posed of a compressive part followed by a tensile part and anothercompressive part due to a moving load pattern (approaching fol-lowed by leaving). Strikingly, the compressive part is unavailablein the IDT fatigue test and could help explain the shorter fatiguelife than the field asphalt concrete because the compressive partplays a beneficial role in the healing of fatigue micro-cracking.
5. Conclusions
Based on this research, the following conclusions can beobtained:
� Compared with the IDT fatigue test, the ABS fatigue testwas more consistent with the tension–compressionresponse of the pavement.
� With the increase in loading cycles, the strain growth rateof the asphalt samples in the ABS fatigue test increasedslower than that of in the IDT fatigue test.
� The fatigue life of the asphalt samples in the ABS fatiguetest was 4–12 times and 12–23 times than that in the IDTtest at 5 Hz and 10 Hz respectively at 20 �C.
� Fatigue damage of the asphalt samples in the IDT fatiguetest developed continuously without recovery but the fati-gue damage recovery was obvious in the ABS fatigue testbecause the creep effect of compressive stress promotedthe healing of micro-cracks in the ABS fatigue test.
On the basis of the results, ABS fatigue test is very promising asan accurate method of evaluating the fatigue life of asphalt mix-tures. However, this current study was limited to the evaluationof laboratory-made asphalt samples. The fatigue performance ofthe materials in the field should be measured by the multiple cores
taken from the pavement next step. Additionally, it is important toquantify healing effect of asphalt samples in the ABS fatigue test.
Acknowledgments
This study was supported by ‘‘the Fundamental Research Fundsfor the Central Universities, SCUT’’ and the open funds for the StateKey Laboratory of Subtropical Building Science in South China Uni-versity of Technology, China (Grant No. 2014KB25).
References
[1] Wang Hainian, Dang Zhengxia, Li Lian, You Zhanping. Analysis on fatigue crackgrowth laws for crumb rubber modified (CRM) asphalt mixture. Constr BuildMater 2013;47:1342–9.
[2] Modarres Amir. Investigating the toughness and fatigue behavior ofconventional and SBS modified asphalt mixes. Constr Build Mater2013;47:218–22.
[3] Li Qiang, Lee Hyun Jong, Kim Tae Woo. A simple fatigue performance model ofasphalt mixtures based on fracture energy. Constr Build Mater2012;27(1):605–11.
[4] Ye Qunshan, Shaopeng Wu, Li Ning. Investigation of the dynamic and fatigueproperties of fiber-modified asphalt mixtures. Int J Fatigue2009;31(10):1598–602.
[5] Zhi Suo, Gun Wong Wing, Hui Luo Xiao, Bo Tian. Evaluation of fatigue crackbehavior in asphalt concrete pavements with different polymer modifiers.Constr Build Mater 2012;27(1):117–25.
[6] Pérez-Jiménez Félix, Valdés Gonzalo A, Botella Ramón, Miró Rodrigo, MartínezAdriana. Approach to fatigue performance using Fénix test for asphaltmixtures. Constr Build Mater 2012;26(1):372–80.
[7] Chen Xingwei, Huang Baoshan. Evaluation of moisture damage in hot mixasphalt using simple performance and superpave indirect tensile tests. ConstrBuild Mater 2008;22(9):1950–62.
[8] Wang H, Al-Qadi IL. Combined effect of moving wheel loading and three-dimensional contact stresses on perpetual pavement responses. Transport ResRec J Transport Res Board 2009;2095:53–61.
[9] Al-Qadia IL, Nassarb Walid N. Fatigue shift factors to predict HMAperformance. Int J Pavement Eng 2003;4(2):69–76.
[10] Adhikari Sanjeev, Shen Shihui, You Zhanping. Evaluation of fatigue models ofhot-mix asphalt through laboratory testing. Transport Res Rec J Transport ResBoard 2009;2127:36–42.
