References: Asphalt Institute. Computer Program DAMA: Pavement Structural Analysis Using Multi-Layered Elastic Theory Users Manual. 1984. Asphalt Research Program, Institute of Transportation Studies, University of California, Berkeley. Fatigue Response of Asphalt-Aggregate Mixes. Strategic Highway Research Program Report SHRP-A-404, National Research Council, Washington, D.C. 1994. Judycki, J. Comparison of Fatigue Criteria for Flexible and Semi-Rigid Pavements. Proceedings, Eighth International Conference on Asphalt Pavements, Seattle Washington. Aug. 10-14, 1997.) Myers, L.A., Roque, R., and Ruth, B.R. Mechanisms of Surface-Initiated Longitudinal Wheel Path Cracks in High-Type Bituminous Pavements. Journal for the Association of Asphalt Paving Technologists. Vol. 67. 1998 pp 401-432.
Historically, fatigue cracking has been considered to start at the bottom of the HMA layer. Fatigue cracks are initiated when the tensile strength of the asphalt concrete is exceeded as the pavement deflects under repeated traffic loads. These cracks continue to grow with increasing numbers of loads. When the cracks are visible on the surface of the pavement, the crack extends the full depth of the HMA layer. Once about 10% of the wheel path exhibits fatigue cracking, it will only be short time before the cracking increases to 45% in the wheel paths.
(Ref: Myers, L.A., Roque, R., and Ruth, B.R. Mechanisms of Surface-Initiated Longitudinal Wheel Path Cracks in High-Type Bituminous Pavements. Journal for the Association of Asphalt Paving Technologists. Vol. 67. 1998 pp 401-432.) This type of fatigue cracking is characterized by longitudinal cracks on one or both sides of the wheel paths. The increase in the use of radial tires (from 80% in 1985 to 98% in 1996) and a corresponding increase in tire pressures of about 20 psi represent a significant change in the surface loading conditions (Florida data). It is these changes in surface conditions that appear to be responsible for increase transverse surface tensions needed to initiate tensile cracking at the tire-pavement interface. This type of fatigue cracking is relatively independent of the type of pavement structure. This indicates that a material’s solution is needed to mitigate this type of cracking.
The most common fatigue test uses a simply supported beam. Dynamic loading is applied to achieve either a constant bending stress or constant bending strain over a number of loading cycles. A cantilevered trapezoidal beam has been used extensively in testing by the University of Nottingham. Similar methods of loading and analysis are used for this testing. Diametral loading can also be used to evaluate fatigue testing since this type of failure is related to the tensile properties of the HMA. However, this method is considered to be less sensitive to mixture variables than beam testing. A notched beam concept has been used with in order to fix the location of the sample failure.
Any of these types of fatigue tests have been used with either dynamic loading or repeated loading methods. The most commonly used is the repeated load.
For controlled-stress testing, failure is defined as the numbers of cycles needed to visibly crack the beam. Failure for the constant strain mode of testing is defined as a 50% loss of initial stiffness. Stiffness at any given cycle is computed from the tensile stress and strain at that specific cycle. The loss of stiffness with numbers of cycles is usually presented as the stiffness ratio (initial stiffness divided by the stiffness at a given cycle). When the ratio is plotted for the test results, it is easy to see when the sample fails (next slide).
The first sample failed after 10,000 loading cycles while the second sample is still above the 50% reduction in stiffness limit at 100,000 cycles.
The Australian fatigue unit is a small, table top device that can test beams using either dynamic or repeated loading. The picture shows a beam loaded in the frame. The clamps at either end hold the beam but are on pivot points that allow the beam to deflect. The two center clamps apply the load and are used to provide a constant moment region in the bottom of the beam (i.e., tensile stress region).
Materials behave differently between constant stress and constant strain loading conditions. When constant stress is used, a stiff asphalt binder will deform very little while a softer asphalt binder will show a much greater deformation (i.e., strain). As conceptually shown in this figure, the strain induced in the softer asphalt binder mix may be very close to the failure strain and would therefore fail faster than the stiffer HMA that is being tested as strains well below failure. This would lead to the conclusion that softer asphalt binders fatigue more quickly.
However, when the same test is conducted with constant strain, the stiffer asphalt binder will be tested at much higher stresses, in some cases close to the failure stress. At the same time the softer asphalt binder mixes have significantly lower stresses at the same strain level. Under these testing conditions, the softer asphalt binder mixes would appear to be the best choice to resist fatigue cracking.
The selection of either constant stress or constant strain should be based on the pavement structure in which the mix will be used. When the HMA layer is less than about 100 mm (4 inches), the mode of failure will be controlled by the large strains at the bottom of the asphalt binder layer. This thickness is typical of low volume roads. When the HMA layer is more than about 150 mm (6 inches) thick, then stress will control the occurrence of fatigue cracking. This thickness is typical of high volume roadways or older pavements that are being overlayed.
Beam and trapezoid beam fatigue testing are similar in many ways. Both simulate flexural stresses seen in pavements but apply uniaxial rather than triaxial stresses. Both reverse stresses (tension-compression) and neither permits the accumulation of permanent deformation with increasing numbers of loading cycles. The only reason for choosing trapezoid rather than beam fatigue testing is the researcher’s preference or local customs. Beam fatigue is typically used in the United States while trapezoid fatigue is popular in the United Kingdom.
This test is simple to perform and uses typical cylindrical specimens. The state of stresses induced in the sample during testing are complex, however, the critical stresses and strains can be calculated assuming linear elastic behavior. A biaxial state of stress is present along the vertical axis with the tensile stress being reasonably constant with significantly more variability in the compressive stress. The main differences between the diametral and beam fatigue tests are that permanent deformation occurs and stress reversal is not practical in diametral testing. Diametral fatigue testing also consistently underestimates the fatigue life relative to other fatigue tests.
This figure gives the student a feel for how the results from each of the fatigue tests compare for the same mix and test method variables.
Laboratory determination of HMA fatigue characteristics takes a considerable amount of time before results can be obtained. Typical testing time for flexural beam fatigue for a set of three samples at each of only two stress (or strain) levels take as long as 3 weeks. In order to obtain estimates of fatigue characteristics quickly, a number of researchers have developed prediction equations for estimating the fatigue life. This section briefly presents some of these predictive equations.
While used on a limited basis for some fatigue studies, researchers at the University of California, Berkeley found that the fracture mechanics approach was excessively complex and difficult. This test, while initially included in the original SHRP research, was eliminated from further study for this reason.
Some of the early research indicated that there may be a unique relationship between the numbers of cycles to failure and the cumulative dissipated energy to failure. That is, the results might be independent of mix variables. Later work indicated that the results were dependent upon mix properties but independent of test methods (two- and three-point bending), temperature (10 to 40 o C [50 to 104 o F]), modes of loading (controlled stress or controlled strain), and the frequency 10 to 50 Hz) (UC Berkeley, 1994).
Conclusions of the testing during the original SHRP program indicated that results were not, as previously thought, independent of the testing variables. Dissipated energy is influenced by the choice of test temperature and the mode of loading. It was found that dissipated energy is highly correlated with incremental decreases in stiffness during fatigue testing which helps explain the effects of mode of loading on mix behavior.
Because of the extensive time required to obtain fatigue results, a number of researchers have developed mathematical models for predicting fatigue from more easily obtainable (in most cases) mix information. This section presents several of the more commonly used equations. The SHRP research resulted in the formulation of an equation for predicting fatigue for actual pavements. The equation constants were calibrated using both field and laboratory data. An evaluation of the SHRP model with a range of mixes indicated that fatigue life, in constant strain, was dependent on aggregate properties such as percent crushing and surface texture. Results indicated that information on the binder alone was not sufficient for predicting actual fatigue life of a pavement. The accumulation of damage is calculated in increments by month. This allows for differences in fatigue cracking due to changing combinations of temperature and traffic loadings.
An evaluation of the SHRP model with a range of mixes indicated that fatigue life, in constant strain, was dependent on aggregate properties such as percent crushing and surface texture. Results indicated that information on the binder alone was not sufficient for predicting actual fatigue life of a pavement.
3HMA Charaterization Fatigue CrackingMechanisms• Traditionally considered to start at thebottom and work up to the top• Crack starts when tensile strain exceedstensile strength of mix• When cracks visible on top, full layercrackedSubgradeBaseAC Mix εtLongitudinal pavement profile
4HMA Charaterization Fatigue CrackingMechanisms• Recent observations of fatiguecracking that starts from the top at theoutside edges of the wheel path• Tensile stresses due to tire-pavementinteractions at surfaceSubgradeBaseAC MixεtTransverse pavement profileTransverse pavement profile
Cantilevered Beam Testing• Trapezoid beamconfiguration• Requires concrete beambe fabricated then sawn• Fixed at bottom, loaded ina cantilever fashion at top
Diametral Fatigue Testing• Repeated load (usually)• Considered less sensitive to mixproperties than flexural
16HMA Charaterization Fatigue CrackingExample of Test Results015,00030,00045,000Cycles toFailure20CTest TemperatureFlexuralTrapezoidDiametralReported in SHRP A-404, 1994
17HMA Charaterization Fatigue CrackingAdvanced Fatigue Topics• Notched-beam test (C* line integral)• Dissipated Energy• Models for Predicting Fatigue Life
Notched Beam Testing• C*-line integral approachFixed Movable
19HMA Charaterization Fatigue CrackingDissipated Energy• Dissipated energy is the amount of energylost for each loading cycle• Calculated from the changes in stressesand strains for each cycle of testing
20HMA Charaterization Fatigue CrackingDifficulties• Research showed that dissipated energyequations are dependent on mix variablesand conditions of testing
21HMA Charaterization Fatigue CrackingPredicting Fatigue fromBinder and Mix Properties• SHRP strain-dependent model• Asphalt Institute’s DAMA Program• University of Nottingham• Shell
22HMA Charaterization Fatigue CrackingSHRP Strain-Dependent Model• Low air voids and crushed, rough-texturedaggregates• Increase stiffness• Increase fatigue life (constant strain)• Indicate that asphalt binder propertyinformation not sufficient for predictingfatigue life