Abstract
The Federal Highway Administration (FHWA) built 12 asphalt pavements in 1993 to validate Superpave tests and specifications used to measure the rutting and fatigue cracking performances of hot-mix asphalts and asphalt binders. Each pavement had four test sites. These sites were tested for either rutting or fatigue cracking using the FHWA’s Accelerated Loading Facility (ALF). The main objective of the study documented in this paper was to compare the fatigue performance results from laboratory bending beam fatigue tests to the ALF fatigue cracking data obtained for these sites from lanes 1 through 4. The four lanes consisted of two asphalt pavement layer thicknesses (100 and 200 mm) and two asphalt binders (PG 58-34 and PG 64-22). Each lane was tested at three temperatures 10, 19, and 28°C. Another objective of this study was to investigate the relationship between the asphalt binder parameter for intermediate temperature performance (G*sinδ) and asphalt pavement fatigue life.Disclaimer
The opinions, findings, and conclusions expressed in this paper are those of the authors only and not necessarily of Jordan University of Science and Technology nor of the Federal Highway Administration.
1.0 Introduction
1 Fatigue failure is known to occur when asphalt pavements undergo repeated loading in the intermediate temperature range from approximately 10°C to 30°C. Significant research has been conducted to understand fatigue cracking and to predict fatigue performance of asphalt pavements. Some researchers, (e.g., Monismith and Deacon, 1969; Van Dijk, 1975 and Kim et al., 1997) characterized fatigue of asphalt mixtures using different theories. Other researchers investigated the factors affecting fatigue response performance. For example, Pell and Cooper, 1975 and Epps and Monismith, 1969 studied the influence of mixture variables on fatigue performance and properties. Raithby and Sterling, 1970; Bonnaure et al., 1982 and Smith and Hesp, 2000 evaluated the effect of rest periods on fatigue performance, characteristics, and fatigue life of bituminous mixtures under repeated loading. Irwin and Gallaway, 1974 and Porter and Kennedy, 1975 referred to the influence of fatigue laboratory test methods on fatigue results for asphaltic concrete.
2 Several laboratory methods and various experimental techniques were cited in the literature to conduct fatigue testing and to characterize fatigue for asphalt paving mixtures. The majority of researchers, (e.g., Monismith and Deacon, 1969; Tayebali et al., 1992, 1993, Carpenter et al., 2003, and Al-Khateeb and Shenoy, 2004) used the bending beam fatigue test set up to conduct fatigue testing for asphalt mixtures. A few researchers including Rowe, 1993 and Kim et al., 1990 used the trapezoidal configuration and the uniaxial configuration for cylindrical specimens, respectively, to test bituminous mixtures for fatigue.
3 Methods and procedures to analysis fatigue testing data varied from research group to another as cited in the literature. Definition of fatigue failure (or the number of load cycles to failure) and determination of the fatigue failure point in fatigue testing has been defined in various ways. Many researchers considered the reduction in initial stiffness in fatigue testing as a measure of fatigue failure. Others observed crack propagation to track fatigue failure and some others introduced dissipated energy concepts to define fatigue failure. Most of the time, the pinning of fatigue failure during a fatigue test was also correlated with the mode of loading. Nevertheless, Al-Khateeb and Shenoy, 2004 developed a fatigue failure criterion that is simple, fundamental, and independent of the mode of loading.
4 This paper documents part of a Federal Highway Administration (FHWA) research study that was performed to assist the highway community in validating Superpave performance-based tests, specifications, and prediction models for asphalt binders and mixtures. Twelve full-scale asphalt pavements wereconstructed in 1993 at the FHWA’s Turner-Fairbank Highway Research Center in McLean, VA, and rutting as well as fatigue performance tests were conducted on these asphalt pavements until 2001 to validate tests for rutting and fatigue cracking. Each pavement had four sites that were tested by an Accelerated Loading Facility (ALF), which is a full-scale pavement-testing machine that applies one-half of a rear truck axle load. The pavements were tested using temperatures, loads, and loading distributions (wheel wander) that promoted either rutting or the formation of fatigue cracks in the asphalt pavement layer.
5 Lanes 1, 2, 3, and 4 were used to conduct the fatigue cracking study, while lanes 5 through 12 were primarily used to conduct the rutting study. The results of rutting studies were also presented elsewhere (Stuart et al., 1999, 2001). The primary objectives of the fatigue cracking study were to validate the Superpave asphalt binder parameter for fatigue cracking, G*sinδ, and to quantify the interacting effects of asphalt binder grade, asphalt pavement layer thickness, and pavement temperature on fatigue cracking performance. G*sinδ is the absolute value of the complex shear modulus times the sine of the phase angle. Another objective of the fatigue study was to validate the strain-controlled bending beam fatigue test developed during the Strategic Highway Research Program.
2.0 ALF Asphalt Pavements
2.1 Pavement Structure and Materials
6 The asphalt pavement structure in lanes 1 through 4 consisted of a single hot-mix asphalt (HMA) on top of an unbound crushed aggregate base and prepared subgrade (Figure 1). The thickness of the asphalt pavement layer was 100 millimeters (mm) in lanes 1 and 2, and 200 mm in lanes 3 and 4. The thickness of the unbound crushed aggregate base layer was 560 mm in lanes 1 and 2, and 460 mm in lanes 3 and 4. The prepared subgrade had a thickness of 610 mm in all lanes.
Figure 1. 1993-2001 ALF Pavements Structure
Display large image of Figure 1
7 The asphalt mixtures in lanes 1 through 4 consisted of two asphalt binders and a single aggregate gradation. Both binders were from Venezuela's Lagoven base stock. Table 1 shows the Superpave performance grades (PGs) of the original asphalt binders (PG 58-34 and PG 64-22) and the continuous PG after the rolling thin-film oven (RTFO) test and after the pressure aging vessel (PAV) test. A diabase aggregate with a 19.0-mm nominal maximum aggregate size was used for the HMA in lanes 1 through 4 with the gradation shown in Figure 2. (Diabase is a dark-gray to black, fine-textured igneous rock composed mainly of feldspar and pyroxene and used for monuments and as crushed stone.) The continuous intermediate temperature PG is defined as the temperature at a G*sinδ value of 5000 kPa and a frequency of 10.0 rad/s. Each mixture should be highly resistant to fatigue cracking at temperatures above its intermediate temperature PG.
Table 1. Asphalt Binders Used in the ALF Pavements
Display large image of Table 1
Table 2. ALF Testing for Fatigue Cracking Resistance
Figure 2. Aggregate Gradation for Each Lane2.2 Pavement Testing Sequence
8 The pavements were tested during October through April when the weather was cool in order to obtain and maintain the required pavement test temperature. The pavements could be heated using infrared heat lamps, but the ALF system did not have the capability of cooling them. Table 2 shows the fatigue testing schedule. Fatigue testing started when the pavements were 1-year old.
2.3 ALF Pavement Testing
9 ALF contains a moving wheel assembly that applies one-half of a rear truck axle load. The ALF machines have long frames of 29 m (95 ft) length with rails to direct rolling wheels. Each ALF machine was capable of applying an average of 35,000 wheel passes per week using a half-axle load ranging from 33 to 84 kN (7,500 to 19,000 lbf). The load was applied in one direction to a 14 m (45 ft) length of pavement at a speed of 18.5 kph (11.6 mph). The machines could allow testing with conventional dual truck tires or wide-based "super-single" tires and simulation of the real-world lateral distribution of truck loadings using programmed transverse wheel wander. To simulate highway traffic, the ALF applies loading on the pavement in one direction only, and the loads can also be distributed from side to side to simulate wheel wander.
10 The ALF was operated using the following protocols:
A minimum of seven distress surveys was performed on each pavement during trafficking. The surveys included measuring transverse and longitudinal profiles and mapping surface cracks. The pavements were cored and trenched after testing. The continuous intermediate temperature PGs of the asphalt binders recovered from the cores are given in Table 3. Age hardening did not affect these PGs as shown in the table. Thus, it was assumed that all of the fatigue beams could be aged using the same protocol. The high- temperature properties of the recovered asphalt binders did show that the asphalt binders aged with time.
Table 3. Continuous Intermediate Temperature PGs of the Asphalt Binders
Display large image of Table 3
11 ALF pavement fatigue life was based on the number of wheel passes when the first crack appeared on the surface of the pavement, and on the number of wheel passes at cumulative crack lengths of 50 m and 100 m. The percent area cracked was also measured. However, this measurement provided the same relative fatigue lives as cumulative crack length. Thus, they are not included in this paper.
2.5 Mechanistic Analysis
12 A mechanistic analysis was conducted using the WinJULEA program to estimate the tensile strains at the bottom of the asphalt layer for each pavement. The inputs needed for each layer were the thickness, modulus, and Poisson’s ratio. The moduli of the crushed aggregate base and subgrade were backcalculated from the data provided by falling weight deflectometer (FWD) tests, which were conducted on the asphalt pavement layer after the pavements were constructed. The program used to do the backcalculations was BOUSDEF. These moduli are given in Table 4(a). Table 4(b) provides the moduli of the asphalt mixtures at the three temperatures (10, 19, and 28°C). These moduli were measured using the indirect tensile test at these three pavement test temperatures. Typical Poisson’s ratios were used as inputs to the WinJULEA program: 0.35 for the asphalt layer, 0.40 for the crushed aggregate base layer, and 0.45 for the subgrade layer. Other inputs were the load of 53 kN, the radius of the loading area, which was 159 mm, and the location where the tensile stresses and strains needed to be determined, which was at the bottom of the asphalt layer. Table 5 provides the tensile strains obtained from the WinJULEA program.
3.0 Laboratory Bending Beam Fatigue Testing
3.1 Sample Preparation
13 The asphalt binders were heated to 163°C and mixed with heated aggregates in proportion to achieve a binder content of 4.8 percent by the total mass of the mixture.
14 The loose asphalt mixtures were short-term oven aged at 135°C for 2 hours according to the AASHTO provisional practice PP2-00 (2000). This degree of aging was found to match the average amount of aging in the pavements prior to the first pavement test. The short-term aged asphalt mixtures were compacted using a Slab-PakTM linear kneading compactor into 180 [.dotmath] 500 [.dotmath] 50-mm slabs. Fatigue beams were then cut from the slabs. The target air-void level for the beams was 7.0 ± 0.5 percent. The air-void range of the pavements was 7.0 ± 1.2 percent.
Table 4. Moduli of the Different Pavement Materials
3.2 Testing Procedure
15 Bending beam fatigue tests were performed in accordance with the AASHTO provisional test method TP8-94 (21). The tests were conducted in the strain-controlled mode of loading. In this test, a vertical sinusoidal displacement is applied to a beam at a frequency of 10 Hz with no rest periods. The required vertical displacement is calculated from the desired peak strain and beam geometry according to Equation (1) shown below. Four strain levels were used in the fatigue testing: 300, 600, 900, and 1200 µε.
Table 5. Tensile Strains at the Bottom of the Asphalt Layer for All Pavements
Display large image of Table 5
16 Two concentrated and symmetrical loads are applied on the fatigue beam specimen, as shown in Figure 3. The beam specimen is forced back to its original position at the end of each load pulse. It is subjected to 4-point bending. Free rotation and horizontal translation are allowed at all load and reaction points.
17 The data acquisition software (TestStarTM) recorded the load and the deformation of the specimen per cycle of loading. The peak tensile stress required to obtain the desired peak strain for each loading cycle was calculated using Equation (2) below:
Display large image of Figure 3
18 The flexural modulus or stiffness (S) was then computed by dividing the peak stress by the applied peak strain as shown in Equation (3).
4.0 Bending Beam Fatigue Test Results
19 In the fatigue test, the modulus (stiffness) of the beam specimen decreases with loading. Since fatigue tests are conducted in the strain-controlled mode (the applied strain is constant), the resulting stress in the specimen decreases with time (or loading). In this study, the most common and widely used fatigue failure method adopted by the AASHTO provisional test method TP8-94 is suggested to define fatigue failure, which is the point at which 50-percent reduction in nitial stiffness occurs. However, in this study, a 40-percent reduction was used because the use of 50 percent would require either extremely long tests or a high number of extrapolations. Thus, a 40-percent reduction in the flexural modulus appeared to be optimal.
20 At a given temperature, each applied strain provides a different number of cycles to failure. The relationship between the applied strain and fatigue life is obtained at each temperature. This relationship is typically used to estimate the fatigue life of an asphalt mixture used in a pavement design. The pavement design provides a tensile strain at the bottom of the asphalt layer using the mechanistic analysis of the pavement. This strain is then used as an input to the relationship to determine the fatigue life. In this study, the strains obtained from the WinJULEA program were used as the inputs.
21 Normally already developed laboratory-to-field shift factors are used to predict the number of equivalent single-axle loads at failure from the bending beam fatigue test results, and to account for differences between the laboratory and the field, such as boundary conditions. Nevertheless, shift factors were not used in this study. The fatigue lives from the bending beam fatigue tests were directly compared to the fatigue lives of the ALF pavements.
22 Table 6 provides the fatigue coefficients K1 and K2 obtained from the developed power model for the relationship between the applied tensile strain and the number of load cycles to failure. The average numbers of cycles at failure based on a 40-percent reduction in the flexural modulus are shown in Table 7. The data using 50- and 30-percent reductions are included as supplemental information. Table 7 also provides the strains actually applied during the bending beam fatigue test. These strains are slightly different from the nominal or desired strain. The relationship between the applied strain and the number of load cycles to failure was developed for the two asphalt mixtures at the three test temperatures (10, 19, and 28°C) using the results tabulated in Table 7 as shown in Figure 4. The developed relationship is a power model of the form shown in Equation (4) below.
23 As noted, the intercept coefficient, K1 and the power coefficient, K2 obtained for the two asphalt mixtures at the three test temperatures are in Table 6.
Table 6. Fatigue Coefficients K1, and K2for PG 58-34 and PG 64-22 Asphalt Mixtures at Test Temperatures of 10, 19, and 28°C
Display large image of Table 6
Table 7. Number of Load Cycles to Failure from the Bending Beam Fatigue Test and the Actual Applied Strains
Figure 4. Beam Fatigue Cycles to Failure vs Applied Strain Level5.0 Fatigue Performance Comparisons
5.1 ALF Pavement Performance vs. Bending Beam Fatigue Test Results
24 The fatigue lives of the asphalt mixtures based on the bending beam fatigue test and ALF pavement performance are shown in Table 8. Table 8(a) groups the pavement data by lane. This shows the effect of test temperature on fatigue life while Table 8(b) groups the data to show the effect of the asphalt binder parameter, G*sinδ, on fatigue life. G*sinδ was obtained from the dynamic shear rheometer (DSR) using the ALF-associated loading frequency of 2.25 rad/s. The asphalt binders were aged using the rolling thin-film oven (RTFO) and pressure-aging vessel (PAV). The bending beam fatigue test results and the fatigue performance of the ALF pavements agree with each other particularly well at the test temperature of 28ºC. Figure 5 shows the relatively high correlation between the two as the R2 values were 0.63 and 0.99 for all test temperatures together and at the test temperature of 28ºC, respectively.
25 For a given asphalt mixture, the best laboratory fatigue performance was obtained at 28ºC and the worse performance was obtained at 10ºC as shown in Figure 4. These findings are to be expected as asphalt mixtures tend to become stiffer at lower intermediate temperatures and therefore their resistance to fatigue cracking becomes lower. A similar trend was obtained for the fatigue performance of the ALF pavements as the correlation between the two was found to be relatively high at the three test temperatures.
26 For ALF lanes constructed using the same asphalt mixture (lanes 1 and 3 or lanes 2 and 4), the asphalt layer thickness did impact the fatigue performance at the ALF sites. For instance, lane 3 with the PG 58-34 asphalt mixture performed better than lane 1 with the same asphalt mixture. Also lane 4 with the PG 64-22 asphalt mixture had better fatigue cracking resistance than lane 2 with the same asphalt mixture. This finding applied to all test temperatures (10, 19, and 28ºC). Thus, regardless of the test temperature, field fatigue performance for thick asphalt pavements was found to be better than that for thin asphalt pavements.
5.2 ALF Pavement Performance vs. Asphalt Binder G*sinδ
27 Figure 6(a) shows the relationship between the ALF wheel passes to initial cracking and the asphalt binder G*sinδ for the pavements with a HMA layer thickness of 100 mm. Figure 6(b) shows similar relationship for the pavements with a HMA layer thickness of 200 mm. As shown in theses figures, a high coefficient of determination (R2) between the ALF fatigue parameter and the asphalt binder parameter (G*sinδ) was obtained particularly for the 100-mm thick pavements. The R2 values obtained were 0.80 and 0.40 for the 100-mm and 200-mm thick pavements, respectively. It has be noted that one or sometimes two outlier(s) were excluded from the analysis due to the fact that this or these data point(s) either corresponded to pavement section(s) that cracked and rutted very quickly because of a local weakness in the crushed aggregate base layer or were obtained using extrapolations.
28 The trend shown in Figure 6(a) indicates that at lower values of G*sinδ, the number of the ALF wheel passes to initial cracking tends to be higher due to the fact that lower G*sinδ values leads to increased flexibility and therefore longer fatigue life. This indicates a strain-controlled response and agrees with the asphalt binder Superpave specifications.
Table 8(a). ALF Wheel Passes to Failure vs. Bending Beam Fatigue Test Results Grouped By Lane
Display large image of Table 8
29 The relationship in Figure 6(b), on the other hand, provides a lower R2 value (0.40) and similar trend to that for the 100-mm thick pavements. In other words, as the G*sinδ value gets higher, the number of the ALF wheel passes to initial cracking becomes lower due to the increased stiffness of the asphalt mixture and thus reduced flexibility.
30 The relationship between the ALF wheel passes to a cumulative crack length of 50 m and G*sinδ for the 100-mm thick asphalt pavements demonstrated in Figure 7(a) having a R2 value of 0.80 was similar to that shown in Figure 6(a) when initial cracking was considered. However, an opposite trend to that shown in Figure 6(b) was obtained between the ALF wheel passes to a cumulative crack length of 50 m and G*sinδ for the 200-mm thick asphalt pavements as illustrated in Figure 7(b) with a R2 value of 0.66. Although higher G*sinδ values results in reduced flexibility, but thicker asphalt layers compensates for the increased stiffness that becomes an advantageous property for fatigue resistance for pavements with thicker asphalt layers over pavements with thinner asphalt layers due to the lower strain levels obtained. In this case, a high G*sinδ value is preferred, and a value such as 2000 kPa should be used as a minimum pass/fail criterion. Below 2000 kPa, there is a wide range of pavement performances which have weak relationship with G*sinδ.
Table 8(b). ALF Wheel Passes to Failure vs. Bending Beam Fatigue Test Results Grouped By Temperature plus Asphalt Binder G*sinδ Data
Display large image of Table 9
31 The proportional relationship between the ALF wheel passes to a cumulative crack length of 50 m and G*sinδ for the 200-mm thick asphalt pavements suggests that the stress-controlled parameter, G*/sinδ, should be used for evaluating the fatigue performance for the thick pavements. Figure 8 shows the relationship using G*/sinδ. The R2 value of this relationship is 0.66. It can be concluded that for the thick pavements, the use of an asphalt binder with a G*/sinδ value below approximately 5000 kPa will provide a greater chance of having poorer fatigue performance. This proves that the response is stress-controlled at all temperatures.
32 For thick asphalt pavements, it cannot be concluded that above a certain strain level, performance will be strain-controlled, and below it, performance will be stress-controlled. For example, thick pavements with strain levels of 422 and 429 microstrains have widely different number of ALF wheel passes to fatigue failure. This means that the fatigue performance of the thicker pavements was not a function of the strain level alone.
Figure 5. ALF Wheel Passes to 50-m Crack Length vs. Beam Fatigue Cycles to Failure
Display large image of Figure 5
Display large image of Figure 6
6.0 Conclusions
33 Based on the results of this study, a number of conclusions may be drawn. After trying several different forms including exponential, logarithmic, and linear, it was found that the relationship between the applied strain and the number of fatigue load cycles to failure was best represented by a power model of the form .
34 For a given asphalt mixture, the best laboratory fatigue performance was obtained at 28ºC and the worse performance was obtained at 10ºC. The bending beam fatigue test results and the fatigue performance of the ALF pavements agreed with each other particularly well at the test temperature of 28ºC although a relatively high correlation was also obtained for all test temperatures together..
35 Increasing the asphalt layer thickness improved the fatigue performance at the ALF sites for lanes constructed using the same asphalt mixture. A good coefficient of determination was obtained for the inverse relationship between the ALF wheel passes to initial cracking and the asphalt binder parameter (G*sinδ) particularly for the thinner100-mm thick pavements. As the G*sinδ value decreased, the number of the ALF wheel passes to initial cracking became higher due to the fact that lower G*sinδ values resulted in increased flexibility and therefore longer fatigue life.
36 A similar relationship was also obtained between the ALF wheel passes to a cumulative crack length of 50 m and G*sinδ for the 100-mm thick asphalt pavements but a direct relationship was obtained between the ALF wheel passes to a cumulative crack length of 50 m and G*sinδ for the 200-mm thick asphalt pavements. This suggests that the stress-controlled parameter, G*/sinδ, should be used for evaluating the fatigue performance for the thick pavements. It could be inferred that for the thick pavements, the use of an asphalt binder with a G*/sinδ value below approximately 5000 kPa would be more likely to exhibit poorer fatigue performance.
Acknowledgement
The authors would like to thank Scott Parobeck and Frank Davis for preparing the asphalt mixtures and fabricating the beam specimens for the bending beam fatigue tests, and Susan Needham for conducting the asphalt binder tests.References
American Association of State Highway and Transportation Officials (AASHTO) (2002) Method for determining the fatigue life of compacted hot-mix asphalt (HMA) subjected to repeated flexural bending. AASHTO Provisional Standards, TP8-94, Washington, D.C.
American Association of State Highway and Transportation Officials (AASHTO) (2000) Standard practice for mixture conditioning of hot-mix asphalt (HMA). AASHTO Provisional Standards, PP2-00, Washington, D.C.
Al-Khateeb, G.G. and A. Shenoy (2004) A distinctive fatigue failure criterion. Journal of the Association of Asphalt Paving Technologists, 62: 344-384.
Bonnaure, F.P., A.H.J.J. Huibers, and A. Boonders (1982) A laboratory investigation of the influence of rest periods on the fatigue characteristics of bituminous mixes. Journal of the Association of Asphalt Paving Technologists, 51: 104-126.
Carpenter, S.H., K.A. Ghuzlan and S. Shen (2003) A fatigue endurance limit for highway and airport pavements. Transportation Research Record, 1832: 131-138.
Epps, J.A. and C.L. Monismith (1969) Influence of mixture variables on the flexural fatigue properties of asphalt concrete. Journal of the Association of Asphalt Paving Technologists, 38: 423-464.
Irwin, L.H. and B.M. Gallaway (1974) Influence of laboratory test method on fatigue results for asphaltic concrete. Fatigue and Dynamic Testing of Bituminous Mixtures, ASTM Special Technical Paper 561: 12-46.
Kim, Y.R., H.J. Lee, and D.N. Little (1997) Fatigue characterization of asphalt concrete using viscoelasticity and continuum damage theory. Journal of the Association of Asphalt Paving Technologists, 66: 520-569.
Monismith, C.L. and J.A. Deacon (1969) Fatigue of asphalt paving mixtures. Journal of Transportation Engineering, 95(2): 317-345.
Pell, P.S. and K.E. Cooper (1975) The effect of testing and mix variables on the fatigue performance of bituminous materials. Journal of the Association of Asphalt Paving Technologists, 44:1-37.
Porter, B.W. and T.W. Kennedy (1975) Comparison of fatigue test methods for asphalt materials. Research Report 183-4, Project 3-9-72-183, Center for Highway Research, University of Texas at Austin, April 1975.
Raithby, K.D. and A.B. Sterling (1970) The effect of rest periods on the fatigue performance of a hot-rolled asphalt under repeated loading. Journal of the Association of Asphalt Paving Technologists, 39: 134-152.
Rowe, G.M. (1993) Performance of asphalt mixtures in the trapezoidal fatigue test. Journal of the Association of Asphalt Paving Technologists, 62: 344-384.
Smith, B.J. and S.A.M. Hesp (2000) Crack pinning in asphalt mastic and concrete: effect of rest periods and polymer modifiers on the fatigue life. Proceedings of 2 nd Eurasphalt & Eurobitume Congress, Barcelona, Spain, II: 539-546.
Stuart, K.D., W.S. Mogawer, and P. Romero (1999) Validation of asphalt binder and mixture tests that measure rutting susceptibility using the accelerated loading facility. Publication No. FHWA-RD-99-204, Federal Highway Administration, McLean, VA..
Stuart, K.D., W.S. Mogawer, and P. Romero (2001) Validation of the superpave asphalt binder fatigue cracking parameter using the FHWA’s accelerated loading. Publication No. FHWARD-01-093, Federal Highway Administration, McLean, VA.
Tayebali, A.A., J.A. Deacon, J.S. Coplantz, and C.L. Monismith (1993) Modeling fatigue response of asphalt-aggregate mixes (1993). Journal of the Association of Asphalt Paving Technologists, 62: 385-421.
Tayebali, A.A., G.M. Rowe, and J.B. Sousa (1992) Fatigue response of asphalt-aggregate mixtures. Journal of the Association of Asphalt Paving Technologists, 61: 333-360.
Van Dijk, W. (1975) Practical fatigue characterization of bituminous mixes. Journal of the Association of Asphalt Paving Technologists, 44: 38-72.