You are on page 1of 8
Prediction of Fatigue Cracking in Asphalt Pavements Do We Follow the Right Approach? A.A. A. Molenaar ‘he fatigue performance of asphalt eonerete pavements is dificult to prediet not only because many of the input parameters needed for the Analyses are difficult to obtain but also because the fatigue phenomenon {tse not well understood. Most analyses, for example, take the tensile strain at the bottom of the asphalt layer as the factor that explains fatigue, although many pavements exhibit top-down eracking that has nothing to do with the tensile strain atthe bottom of the asphalt layer. ‘This approach implies that calibrating fatigue predictions onthe basis of tensile strains atthe bottom of the asphalt layer with the amount of cracking observed at the pavement surface is rather unrealistic, Further- more, Its unlikely that fatigue eracks, i initiated atthe bottom of the asphalt ayer, show up at the pavement surface as clearly defined eracks. ‘There are many indications tha fatigue at the bottom of the asphalt layer {sa matter of the development of a deteriorated zone with microcracks rather than development of clearly defined, diserete cracks. Itis ‘that large shift Factors are needed to apply laboratory futigue relations to field predictions. The magnitude ofthese shift factors depends, among ‘other determinants, on the type of fatigue test and mode of loading, AS this paper, only the slope of the fatigue relation ean be est ‘mated with confidence from laboratory fatigue tests. Finally, this paper shows that the existence ofan endurance limitcan be debated; inany case itisnot a constant value of about 70 um/m, Experiments have shown that the suggested value Is too high. nt For many years fatigue cracking in asphalt pavements has been a sub- ject for research, Traditionally itis belioved that fatigue cracks are a result of tensile strains at the bottom of the asphalt layer. These ten sile strains result in cracking, orin more general terms, damage, which propagates upward and becomes visible at the pavement surface as Tongitudinal or transverse cracks in the wheel tracks, resulting in alli ‘gator cracking. The problem with the prediction of fatigue cracking is thata significant shift in the Iaboratory-determined fatigue relation has to be applied to obtain a so-called field fatigue relation. This shift is ‘caused by effects such as lateral wander of traffic, healing, and differ- ences in geometry and test conditions in pavements when compared with test setups. ‘The required calibration factors have to be derived by matching results of performance predictions to field performance. As shown in this paper, matching field data to experimental results is not asim- Faculty of Gil Enginesing and Geo Sclences, Dele Unversity of Tec P.O. Box 5048, 2800 GA Delt, Netherlands. ..0.molenear@cita tude Transportation Research Recor: Joumal of the Transportatian Research Board, No. 2001, Transportation Research Boardaf te National Academie, Washington, 0.C., 2007, pp, 158-162, Do; 10.3141/2001-17 185 ple task, because most of the observed wheelpath cracking might well be top-down cracking, which has little to do with the “classical” bottom-up type of fatigue cracking, Furthermore it has to be determined whether fatigue damage ini ‘iating atthe bottom of the asphalt layer really is propagating upward as discrete cracks that are so nicely visible at the pavement surface. ‘As shown in this paper, itis more likely that a zone of deterioration develops at the bottom of the asphalt Iayer. This zone will exhibit ‘microcracks and will result in « higher overall strain level atthe bot tom of the asphalt layer as well as a reduced effective stiffness of the layer. Finally, the question has tobe raised, does asphalt concrete exhibit, just like portland cement concrete, an endurance limit, mesning that below a certain strain limit, fatigue will no longer be a problem? Al of these aspects have a significant effect on how to predict fatigue performance of asphalt pavements and are discussed in this paper. FATIGUE PERFORMANCE OF PAVEMENTS Figure 1 shows an example of fatigue cracking asitcan be observed in the field, Wheel track cracking is a combination of longitudinal, transversal, and some alligator cracking. Similar patterns have been observed on test sections such as those of the Delft University accelerated pavement testing device Lintrack (Figure 2). ‘Testing of a 150-mm-thick asphalt section (Iwo layers of gravel asphalt concrete, top layer 70-mmm thick, 4.6% voids, 4% bitumen bby mass, density 2,371 kg/m?, bitumen PEN 43, Tyan 53°C) that ‘was directly placed on a sand subgrade showed that many of the cracks were present only in the topmost 50 mm of the pavement, ‘whereas some cracks had propagated from the bottom upward. Fur- thermore cores revealed that the lower half ofthe asphalt layer was seriously disintegrated around the center of the trafic line. In fact ‘at some locations the material emerged as an amount of loose rub- ble, some of the coarse aggregate being completely stripped. Hav- ing top-down cracking as the dominating failure mode, the question arose about whether classical bottom-up fatigue had occurred or not. This question could be answered by analyzing the results of strain gauge measurements and falling weight deflectometer results. Because the tensile strain at the bottom of the asphalt layer was ‘measured at various locations and because falling weight deflectome- {er measurements were done at regular time intervals, the increase in tensile strain during the accelerated pavement testing (APT) as well 1s the decrease in effective stiffness could be determined. Examples of such results are shown in Figures 3 and 4 (1). Both figures clearly 156 FIGURE 1 Wheel track cracking due to fatigue. show that damage is developing atthe bottom of the asphalt layer and thatthe effective stiffness is decreasing, Furthermore approxi rately 20% of the trafficked erea was cracked when the effective modulus, back-calculated from falling weight deflectometer tests performed in the wheel track, had reduced to 50% of its original value, It should however be kept in mind that most ofthe observed cracking was surface cracking! 420 - ‘Transportation Resesrch Record 2001 A similar tendency was observed on a test section in which a '50-mm-thick layer of the same asphalt mixture was placed on the same subgrade. Also in this section the moment at which the asphalt stiffness had reduced to approximately 50% ofits initial value co- incided with the moment at which approximately 20% of the wheel track showed cracking, In tis case it was difficult to decide whether the cracking was top down or bottom up because most of the cores taken on a crack showed that the crack had developed through the entire asphalt layer. ‘These observations lead to the conclusion that bottom-up fatigue 2p 56x25| Z ol z eee Fo} mao ST tet 70550 or = senso gm Boor See . e 2 | |——2mnss S g 20 8 06 1) _ pnt 70325 | > = | | 2nt 7080 ‘6 ! | os 40 60 80 100 0 20 40 60 80 100 Nt. of repens Neo ropettons @) Q) FIGURE 10 _{a) Increase in phase angle and (6) decrease of applied force in relation to number af loed repetitions (horizontal exis) Molenaer 181 Fatigue relationships Lintrack sections 288 . as + hoe B, 4 ie ate , “ © orc 55 }—~ % © 1,20 5 e = Mv,ore > liv, 20% 4 8 + regression °C 4 regression 20°¢ 3s Be 15175 2 22528 tog (strain avn) FIGURE 11. Fatigue results obtained on gravel asphalt concrete base course 85) is too optimistic. From the discussion presented above, itis clear that the concept of “endurance limit” may be valid for asphalt pavements, but the sug- gested threshold tensile strain value of 70 ym/m is certainly not a Constant value and certainly too high for many mixtures, ‘TOP-DOWN CRACKING tis well-known that the contact pressure distribution is far more complex than usually assumed in design analyses, The contact [pressures are highly nonuniform and are acting not only in the ver~ tical but also in the horizontal (longitudinal and transversal) direc tion. Van Gurp and Wennink (71) present results for calculations to determine the tensile transversal strain near the edge of the tire, which isa result of the nonuniform transversal shear stresses act- ing undera 50-KN super single tire and the tensile strain at the bot- tom of the asphalt layer. Furthermore these tensile strains were correlated with the bending stiffness of the pavement as measured bby means of a falling weight deflectometer. Equation 6 predicts the tensile strain atthe pavement surface, which can be held responsible for top-down cracking. 4, =194.9 = 20.78 logSChyy o ‘where 6 equals tensile strain atthe pavement surface (uv) due to the 50-KN super single tre and SClya equals the maximum deflection ‘measured with flling weight deflectometer (F = 50 kN)—deftection ‘measured at a distance of 300 mm (im). Equation 7 predicts the ‘onsie strain al the bottom ofthe asphalt layer). log €= 0.481 + 0.881 1ogSCly, o ‘where € equals the tensile strain at the bottom of the asphalt layer (jan/m) resulting from a 50-KN super single tir. Equations 6 and 7 imply thatthe tensile strain atthe top of the asphalt layer becomes dominant at SC» values of about 90 ym and lower. This implies that especially for pavements with a fairly high bending stiffness, surface fatigue cracking becomes dominant over ture showing that proposal of endurance limit of 70 ym/m (logarithmic fatigue cracking thats initiated atthe bottom of the asphalt ayer. That isthe case, or example, for a pavement consisting ofa 150-mm-thick asphalt top layer with & modulus of 6,000 MPa on a 300-mm-thick ‘ase with a modulus of 300 MPa on a subgrade with a modulus of 100 MPa, Because structures with a similar bending stiffness are ‘quite common for main highways as well as the secondary road system, itis most likely that top-down eracking is dominant on such roads. HOW TO PROCEED From the discussion presented so far, the following conclusions can be drawn: 1, Laboratory fatigue tests are providing only specimen properties rather than mixture properties. 2. ‘The only material property that can be obtained from laboratory fatigue testing is the slope of the fatigue line. However, this paper indicates that the value for this parameter can be obtained in a much -more practical way by means of resilient modulus testing. 3, Shift factors of significant magnitude need to be applied on laboratory fatigue relations to be able to use them for field fatigue performance predictions. 4, In many cases, top-down cracking might be dominant over bottom-up cracking, This phenomenon should be taken into account inpaverent design procedures, because it causesa significant amount of maintenance. 5. Calibration of fatigue relations by matching analysis results to cracking observed at the pavement surface might result in erroneous calibration factors, ‘These conclusions imply that the prediction of fatigue perfor- ‘mance is more an art than a science. By using the currently available techniques combined with a great deal of practical experience, it appears tha it is possible to compare different structures and materi als with each other. An actual prediction of the fatigue life is not pos- sible, given the uncertainties around the magnitude of the shift and 162 calibration factors that need to be applied to match laboratory fatigue \with field fatigue performance. These factors might differ from one location tothe other. ‘The fact that our current procedures are not good enough is clearly demonstrated by De La Roche etal. (12), who show that the ranking ofthe fatigue performance of mixtures when tested in the French APT device Manege de Fatigue was not the same asthe ranking determined bby means of two-point beam bending fatigue tests even when the effect of rest periods was taken into account, ‘This clearly indicates the need for improved testing and modeling ‘tis quite clear that fatigue characterization, iocluding healing, should bbe done in a way that is test independent, Furthermore analysis tech- niques are needed that allow damage initiation and propagation pre- dictions to be made including, among othess, the effects of stress redistribution. Nice examples of such techniques based on an elas- toviscoplastic approach are presented, for example, by Erkens (/3) and Medani (14). Unfortunately space does not allow a detailed discussion ofthese models, REFERENCES: 1, Groenendij, 1. Accelerated Testing and Surface Cracking of Asphaltc Concrete Pavements. PhD dissertation, Delft University of Technology, Delft, Netherlands, 1998. 2, Pronk, A.C. Harmonisation of Bending Fatigue Tests: A(n) (mpossibi (in Dutch). Proc., Weghoukundige Werkdagen 1998. CROW, Ede, Netherlands, 1998. 3. Van Alphen, C. A.M, and A.A.A. Molenaar, Fatigue Tests on an Asphalt Mixture Using the 4 Poin Bending and Indirect Tension Tet (12 Dich). Report 7-85-113-8, Road and Railways Research Laboratory, Delft University of Technology, Dei, Netherlands, 1985 4, Huurman, M, and M. van de Ven, Enrope end the Benefits of Modelling ‘Test Setups, Measurements Don’t Always Tell What You Think Dutch). In Wegbounkundge Werkdagen 2004 (CD-ROM). CROW, Ede, Netherlands, 2004 ‘Transportation Research ecard 2001 5, Collop,A.C.,A. Searpas, C. Karsbergen, and A. de Bondl, Development and Finite Element Implementation of Stress-Dependent Elastoviseo- Plastic Constitutive Model with Damage for Asphalt. In Transporta- ton Research Record: Journal ofthe Transportation Research Board, ‘No. 1832, Transportation Research of the National Academies, Wash ington, D.C, 2003, pp. 96-104. 6. Jacobs, M. M.J. Crack Growl in Asphaltic Mics, PAD diss. Del Unt versity of Technology, Delt, Netherlands, 1995, 7, Molenaar, A.A. A. Structural Performance and Design of Flexible Pavements and Asphalt Conerete Overlays. PAD dissertation, Delt Unie versity of Technology, Delt, Netherlands, 1983, ‘8, Medani,T.O.,and A.A. A. Molenaat, A Simplified Practical Procedure for Estimation of Fatigue and Crack Growth Characteristics of Asphaltic ‘Mixes. Intemational Journal of Road Materials and Pavement Design, Vol, 1, No.4, 2000. 9, Ayets, M. E. Status of the Mechanistio-Empirical Pavement Design Guide (M-EPDG). Technical services memo. American Concrete Pavement Association, Washington, D.C., 2006, 10. Molenasr, J.M.M. Performance Related Characterisation of the Mechanical Behaviour of Asphalt Mixtures. PhD dissertation. Delft University of Technology, Delt, Nethertands, 2008, 11. VanGurp,C. A.P.M. and P.M. Wennink. Rural Road Pavement Design ‘and Evaluation in Dutch). Report E96007, KOAC-WMDP Consultants, ‘Apeldoor, Netheriands, 1996, 12, DeLaRocke, C, H. Odeon, J-P. Simoncelli and A. Sperm, Stwdy ofthe Fatigue of Asphalt Mixes Using the Crular Test Track ofthe Labora- loire Central des Ponts et Chaussees in Nantes, France. In Transportation Research Record 1436, TRB, National Research Council, Washington, DC, 1994, pp. 17-21. 13. Bikens,S. M.J. G. Asphalt Concrete Response (ACRe}~Determination, ‘Modelling and Prediction. PAD dissertation. Delft University of Technol- ‘ogy, Deft, Netherlands, 2002, 14, Medani, TO. Design Principles of Surfacings on Orthotropic Steel Bridge Decks. PhD dissertation, Delft University of Technology, Delft, Netherlands, 2006. Tho Characteisties of Bituminous Paving Micures to Meet Structural Requre- ‘ments Committee sponsored pulioaton of this paper.

You might also like