A Simplified Practical Procedure for Estimation of Fatigue and Crack Growth Characteristics of Asphaltic Mixes

T.O. Medani* and A. A. A. Molenaar** * Research Engineer, ** Prof. of Highway Engineering Faculty of Civil Engineering and Geo Sciences P.O. Box 5048 2600 GA Delft The Netherlands t.medani@citg.tudelft.nl a.molenaar@citg.tudelft.nl
ABSTRACT: This paper describes a simplified procedure for estimation of fatigue and crack growth characteristics of asphaltic mixes. The procedure requires the determination of the socalled master curve(s), the asphalt properties and the mix composition. It is shown that the master curve does not only give information about the variation of the mix stiffness, the loading time and the temperature, but also on the fatigue behaviour of asphaltic mixes. This information can reliably be obtained from rather simple tests, which can be performed in many road-engineering laboratories. In the proposed procedure it is possible to choose between the Whler approach and the more fundamental fracture mechanics approach. Procedures of determining the fatigue characteristics of asphaltic mixes from simple tests are of special interest in mix design procedures where the quality of various mixes with respect to fatigue and crack resistance has to be evaluated. It is obvious that such a procedure is also extremely attractive for specification and pavement evaluation purposes. An insight is provided as to how the fatigue parameters for an asphaltic mix can be estimated using the Whler approach and the fracture mechanics approach. KEY WORDS:

fatigue, crack growth, simple tests, Whler approach, fracture mechanics

International Journal of Road Materials and Pavement Design, Volume X, No. X/2000

1. Introduction In most European countries, the amount of traffic is increasing beyond all expectations. Transport vehicles are increasing not only in number, but also in magnitude. The use of super single tyres and different axle configurations is increasing too. The effects of these factors tend towards increased pavement deterioration, including fatigue, resulting in increased costs to maintain road networks to an adequate standard. Furthermore, the need to predict the remaining life of asphalt pavements and the design of new pavements to withstand heavier traffic loading with new axle and suspension configurations requires knowledge of the critical asphalt strain levels. This necessitates knowledge of the mechanical properties of the materials and the fatigue characteristics of asphalt mixtures. Fatigue cracking in asphalt pavement layers arises from repeated tensile strains due to traffic loading. In the design procedures, which are commonly in use, the maximum tensile strain is calculated at the bottom of the bituminous layer. This implies that fatigue cracking initiates at the bottom of the asphalt and propagates upwards. It is very difficult to verify this hypothesis in practice. However, strain measurements in sections tested at the Delft University accelerated pavement testing device, LINTRACK, indicted the presence and growth of cracks at the bottom of the asphalt layer [GRO 99]. Furthermore, work done by van Dijk [DIJ 75] on laboratory produced slabs tested by means of a small scale wheel tracking device has shown the hypothesis to be true. Nevertheless, most of the cracks that are observed in practice initiated at the pavement surface and have propagated downwards. This phenomenon has already been investigated in the early and mid 80s of the past century and it has been shown [MOL 83; GER 88] that these surface cracks are most likely due to horizontal shear stresses in the tyre-pavement interface. These shear stresses can result in local high tensile strains near the tyre edge and can be held responsible for longitudinal cracking near the edges of the tyre. The development of these longitudinal surface cracks is accelerated when ageing of the wearing course occurs. The assessment of fatigue characteristics is normally done through fatigue tests, but these tests are not possible in many cases, either they are not available because they are too costly, or time consuming. The only alternative till now is the use of nomographs. However, these nomographs have been developed from results obtained out of tests performed under specific conditions. For that reason it seems extremely useful if the estimation of the fatigue characteristics can be made from results of relatively simple and inexpensive tests which can be performed in any road-testing laboratory.

Estimation of fatigue characteristics of asphaltic mixes using simple tests

Procedures of determining the fatigue characteristics of asphaltic mixes from simple tests are of special interest in mix design procedures where the quality of various mixes with respect to fatigue and crack resistance has to be evaluated. It is obvious that such a procedure is also extremely attractive for specification and pavement evaluation purposes.

2.

Fatigue damage models

The fatigue damage models that are used in asphalt road engineering are the Whler approach, the dissipated energy approximation and the fracture mechanics technique.

2.1. Fatigue relationship When an asphalt beam is subjected to repeated flexure (e.g. four-point bending test) the maximum flexural tensile strain occurs at the outer fibre of the beam. The number of load repetitions to failure is related to the tensile strain following a Whler-type fatigue relationship:

1 N f = k1
where: Nf

[1]

: number of strain applications to failure, : strain at the bottom of the asphalt layer and : factors, depending on the composition and properties of asphalt mix. the

k1, n

Using the fatigue relationship (Equation 1) as base for the description of fatigue results, several researchers found material depending characteristics for the parameters k1 and n. Some of the relationships, determined by regression analysis, are: SPDM [SPD 78]:

1 N f , SPDM = (0.856Vb + 1.08) 5 (10 6.S m ) 1.8 ( ) 5

[2]

International Journal of Road Materials and Pavement Design, Volume X, No. X/2000

Bonnaure et al. [BON 80]:

1 N f , Bon = (4.402 PI 0.205PI .Vb 2.707) 5 (10 6.S m ) 1.8 ( ) 5

[3]

N f , SHRP = 2.738.10 5 exp 0.077VFB 3.624 (S m sin )

Nf

SHRP [SHR 94]:

2.720

[4]

where: :load cycles to failure, :strain amplitude, Vb :volume percentage of binder in the mix, Va :volume percentage of air in the mix, Vg :volume percentage of aggregate in the mix, PI :penetration index, Sm :stiffness modulus (in MPa), pen :penetration of the bitumen (in 0.10 mm), VFB :voids in the aggregate skeleton filled with bitumen = Vb/( Vb+ Va) and :phase angle It appeared that the phase angle is highly correlated with the stiffness of the mixture [psi] following:

= 260.096 17.172 ln S m

[5]

It can be noted that in equations [2], [3] and [4] that the value of the slope of the fatigue relation, n, takes a constant value of 5 in equations [2] and [3], and 3.624 in equation [4]. This assumption is not completely correct. Thats because it would mean that at low temperatures and short loading times, when the material is brittle, a same value would be valid as at high temperatures and longer loading times when the material is more flexible. The brittle behaviour should be represented by a high n value, while a flexible behaviour should be represented with a low n value [MOL 00].

2.2. The dissipated energy approach The dissipated energy is an alternative approach to fatigue characterisation, irrespective of the loading conditions during the test. Van Dijk [DIJ 75] showed that the total energy dissipated per volume to fatigue failure can be written as:

Estimation of fatigue characteristics of asphaltic mixes using simple tests

5 [6]

Wf = Bf N z f
where: Wf Nf Bf, z : total amount of dissipated energy per volume (J/m3), : number of load repetitions to fatigue failure and : constants

According to Van Dijk and Visser [DIJ 77] this relationship is independent of: the type of test method, loading frequency (between 10 and 50 Hz), mode of loading, rest periods (between 0 and 5 times the loading time), temperature (between 10 and 40oC) and the type of bitumen (40/50-80/100).

2.3. The fracture mechanics approach Notably Paris and Erdogan [PAR 63] and Rice [RIC 68] have developed fracture mechanics for bituminous materials from theory originating in the field of metallurgy. In the fracture mechanics approach three stages are distinguished (e.g. Ewalds and Wanhil [EWA 86]): 1. the initiation phase: development of hairline or micro cracks, 2. the propagation phase: development of macro cracks out of micro cracks, 3. The disintegration phase leading to collapse and initial failure of the material. In this phase unstable crack growth can be observed. Paris and Erdogan [PAR 63] first described the crack propagation phase. They found, by analysing experimental data using regression analysis, for repetitive loading conditions that the crack propagation rate dc/dN is:

dc = AK n dN
where: A, n

[7]

: parameters, depending on the material and on the experimental conditions (waveform, temperature, frequency),

International Journal of Road Materials and Pavement Design, Volume X, No. X/2000

c N K

: crack length (mm), : number of load repetitions and : stress intensity factor.

3. Estimation of fatigue and crack growth characteristics of asphaltic mixes using simple tests Repeated loading fatigue tests are costly and time consuming. Therefore a methodology was developed at the Delft University of Technology mainly by Jacobs [JAC 95], extending the work of others, to derive the fatigue characteristics of asphaltic concrete (AC) materials from simple, relatively fast and inexpensive tests. This methodology is based on fracture mechanics and uses Paris law which describes the crack growth process. The methodology to derive the fatigue characteristics of AC materials consists of two main steps: the first step is the derivation of the crack growth characteristics of the material (i.e., the parameters A and n of Paris law). The second step is the application of these crack growth characteristics in modelling of the fatigue process as the growth of one equivalent crack. This step requires the input of the actual geometry of the fatigue process being modelled (four-point bending test, or actual pavements, etc). In short, in its present form, the methodology requires three main inputs: 1. 2. 3. a master curve of log mix stiffness versus log loading time, a tensile strength of the AC mix as a function of bitumen stiffness, fracture energy of the mix as a function of bitumen stiffness.

Jacobs methodology still requires conducting uniaxial static tensile tests to determine the master curves for tensile strength and the fracture energy, backcalculation of bitumen stiffness from mix stiffness. He further uses some regression constants developed for certain mixes. If the fatigue characterisation is desired for a mix that was not tested by Jacobs then the average constants for his mixes might be used. These constants will be used in the determination of many variables required as input for his procedure. This might lead to accumulation of errors. Therefor, it is thought that a more simplified and fast method will be extremely useful for the estimation of fatigue and crack growth characteristics. Such a procedure is of special interest in mix design procedures where the quality of various mixes with respect to fatigue and crack resistance has to be evaluated. Furthermore, it is also extremely attractive for specification and pavement evaluation purposes.

Estimation of fatigue characteristics of asphaltic mixes using simple tests

4. A proposed simplified procedure for estimation of fatigue and crack growth characteristics The proposed modified version is simpler, more flexible and can be used for practical purposes for the characterisation of the fatigue and crack growth parameters. The procedure allows a choice between the famous phenomenological fatigue equation and the fracture mechanics approach. Both approaches require the determination of n from master curves derived from e.g. simple dynamic indirect tensile tests. Then the k1 parameter in the Whler-type fatigue formula or the A in Paris' law can be estimated using regression equations.

4.1. Estimation of n Work carried at the Road and Railroad Research Laboratory (RRRL) of the Delft University of Technology [MOL 83; JAC 95; SAB 95; MED 99] have shown that the slope of the fatigue relation is dependent on the slope of the relationship between the mix stiffness and the loading time. The relationship between the mix stiffness Sm and the loading time t at a certain reference temperature can be represented by the log-log equation:

log S m = f (log t )
The slope of the master curve m at a certain loading time can be determined as:

[8]

m=

d (log S m ) d (log t )

[9]

An example of such a relationship is shown in Fig. 1. Schapery [ SCH 73, SCH 75A, SCH 75B] derived the following theoretical relationship between n and m (strain-controlled condition) as:

nmas =

2 m

[10]

International Journal of Road Materials and Pavement Design, Volume X, No. X/2000

where: nmas :n-value determined from the master curve

100000

10000

(Sm)[Mpa]
1000 100 0 .0 1

0 .1

10

100

t (s )

Figure 1. Example of the relationship between log Sm and log t However, differences between the n-values as estimated from the master curve (nmas) and those determined from the four-point bending test (nexp) were reported [MOL 83; JAC 95; MED 99]. This difference can be attributed to the limitations of the model as Schapery's theory is developed for ideal visco-elastic material. This assumption is not fully applicable to asphaltic mixes since they contain voids and aggregate particles which influence the crack growth. A correction factor (CF) is then defined as:

CF =

n mas n

[11]

The regression equation found for the correction factor is:

CF = 0.541 + 0.173nmas 0.03524Va

[12]

Estimation of fatigue characteristics of asphaltic mixes using simple tests

with R2 equals to 0.92 and a standard error of the estimate of 0.25. In Figure 2 it is shown how the n values that were determined experimentally for 38 mixtures, nexp, agreed with the n values that were estimated from nmas and the correction factor, nest. It can be concluded that a very good agreement is obtained.

35

nmas
30

nexp
25

nest

20

n
15 10 5 0 0 5 10 15 20 25 30 35 40

experiment number

Figure 2. Comparison of n-values obtained from experiments, master curves and estimated by means of the correction procedure Slightly higher values for R2 could be realised if the results obtained for some mixtures were excluded, but equation [12] is preferred for practical reasons. That is because 10 different types of mixes including modified (SBS modification) and nonmodified mixes were included in the derivation of the equation, and hence it is preferred for the more comprehensive nature of its applicability. In Table 1 an overview is given of the range in volumetric composition of the mixtures that were involved to obtain the relationship to estimate the exponent n of the displacement controlled fatigue test.

10

International Journal of Road Materials and Pavement Design, Volume X, No. X/2000

Vb % 11.0-20.7

V a% 1.0-8.3

Vg% 71.0-89.6

Note: Vg= volume percentage of aggregates

Table 1. Range of mix compositions involved After the determination of nmas from the master curve the n-value can be directly estimated. In order to characterise the fatigue behaviour of the asphaltic mix the k1 component of the fatigue equation is needed. If it is desired to use the fracture mechanics to analyse the crack growth characteristics then the A-parameter of Paris' law needs to be estimated. In the following parts the estimation of these parameters will be discussed. 4.1.1. Estimation of the fatigue equation parameter log k1 Regression analysis performed on the results of 108 displacement controlled fatigue tests to derive a relationship between the constant k1 and the composition of the asphalt mixtures resulted in the following equation:

log k1 = 6.589 3.762n +

V 3209 + 2.332 logVb + 0.149 b + 0.928PI 0.0721 R&B [13] T Va Sm

with R2 =0.985 and a standard error of the estimate of 0.595. where: n, log k1 Vb Va PI Sm TR&B : the fatigue equation parameters : volume percent of bitumen : volume percent of voids : penetration index of bitumen : mix stiffness (MPa) : ring and ball temperature of bitumen (oC)

The data used to derive this relationship were taken from fatigue tests performed by SHELL, the Road and Hydraulics Engineering Division of the Dutch Ministry of Transport, and the Road and Railways Research Laboratory of the Delft University of Technology. The tests were either 3 point or 4 point bending fatigue

Estimation of fatigue characteristics of asphaltic mixes using simple tests

11

tests. The mix composition, bitumen properties, test conditions and fatigue parameters for the data set used is shown in Table 2.

Vb[%] 6.1-19.3

Va [%] 1.9-30.9

Vg [%] 69-88.10

f [Hz] 10-50

T [OC] -10- +35

TR&B [OC] 43-78

PI -1.5-2.0

Table 2. Composition of the mixtures involved in development of the equation for k1 Figure 3 shows the agreement between the estimated and the experimentally determined log k1 values. It is believed that the proposed equation is capable of providing very good estimates for log k1 as only 3 out of 108 points lie outside the range of 10% and all the points are predicted within the range of 10 to 14%.

Figure 3. Comparison between the experimentally determined and predicted log k1

The quality of the proposed relationship was tested by predicting the log k1 values of 10 mixtures that were tested in the SHRP-A-404 Project [SHR 94]. It appeared that the difference between the predicted and experimentally determined

12

International Journal of Road Materials and Pavement Design, Volume X, No. X/2000

log k1 values remained between 6% and +9%, which is the same range as found for the data from which the equation was developed. 4.1.2. Estimation of the A- parameter of Paris' law

In order to use the fracture mechanics for pavements design purposes the crack growth characteristic A of Paris' law should be determined. Again regression analysis performed on the results of 192 tests conducted by Molenaar [MOL 83] and Jacobs [JAC 95] to derive a relationship between A and n as:

log A est = 2 . 890 0 . 308 n 0 . 739 n 0 . 273 log S mix

with R2 of 0.888 and a standard error of the estimate of 0.58. Table (3) gives the range of composition of the involved mixes.

[14]

Vb 5.7-16.7

Va 1.30-20.5

Vg 69.9-57.2

Table 3. Composition of mixes involved in development of the equation for A The proposed equation gives a reasonable estimate for log A as 92% of the results are predicted within 10% to 25%. This is shown in Figure 5.

Estimation of fatigue characteristics of asphaltic mixes using simple tests

13

Figure 5. Comparison between the experimentally determined and predicted log A

Equation (14) can be compared by similar equations developed by Molenaar [MOL 83] and Jacobs [JAC 95]. Molenaar [MOL 83] proposed the following equation with R2 of 0.884:

log Aest = 1.628n 0.977

Jacobs [JAC 95] proposed the following equation with R2 of 0.83:

[15]

log Aest = 1.14n 2.36

[16]

However, the data set used for the derivation of equation (14) is larger than the set data used by Molenaar [MOL 83] and Jacobs [JAC 95] and the R2 is slightly higher. Probably, equation (14) can be considered as a practical formula for the estimation of A.

14

International Journal of Road Materials and Pavement Design, Volume X, No. X/2000

4.

Stepwise description of estimation of the fatigue parameters for an asphaltic mix

In order to determine the fatigue relationship for a specific mix for a given temperature and a particular loading time the following stepwise procedure is described: -step 1 Prepare specimens for the determination of the master curve for the stiffness modulus by means of the repeated load indirect tensile test set-up. The dimensions of the specimens should be chosen such that the assumption of plane stress conditions holds. According to Lytton et al. [LYT 93] the thickness of the specimen should not be over 30 mm for a 102-mm diameter specimen. -step 2 Determine the relationship between the stiffness modulus and the loading time at the temperature of interest (master curve) using, e.g. a repeated load resilient modulus test. After constructing the master curve regression analysis is used to define the relation:

log S m = f (log t )
-step 3 Calculate nmas using:

[17]

nmas =

2 2 = m d (log S m ) d (log t )

[18]

-step 4 Estimate nest from:

Estimation of fatigue characteristics of asphaltic mixes using simple tests

15

nest =

n mas CF
and

[19]

CF = 0.541 + 0.173nmas 0.03524Va

[20]

-step 5 If it is intended to use the fatigue relationship then the k1-value is estimated from the mix composition and the bitumen properties as:

log k1 = 6.589 3.762n +

V 3209 + 2.332 logVb + 0.149 b + 0.928PI 0.0721TR& B [21] Sm Va

If it is desired to use Paris' law as a method to describe the crack growth in elastic media then the A value can be estimated from:
Sm

log A est = 2 . 890 0 . 308 n 0 . 739 n 0 . 273 log

where all symbols are as previously defined

[22]

6. Conclusions Based on the material presented in this paper, the following conclusions have been drawn: - It has been shown that the relationship between the loading time and the stiffness modulus, the so called master curve, can be established by e.g. repeated load indirect tensile tests which can be performed easily at relatively low. - A simplified procedure for fatigue and crack growth characterisation of asphaltic mixes is proposed. The procedure requires the determination of the master curve as the only input.

16

International Journal of Road Materials and Pavement Design, Volume X, No. X/2000

- The equation to predict the slope of the fatigue line n has been developed from 38 fatigue tests including tests on polymer modified mixtures. - A relationship between the k1 parameter of the fatigue equation and n (which can be estimated from the master curve), volumetric composition and bitumen characteristics has been proposed. This equation allows the characterisation of the fatigue behaviour of asphaltic mixes using the phenomenological Whler equation. - The equation to predict k1 has been developed using the results of 108 fatigue tests. The equation has successfully been used to predict the k1 values of 10 asphalt mixtures that were tested in the Strategic Highway Research Program. - A relationship between the A -parameter of Paris' law and n estimated from master curve was proposed. This relationship allows the application of Paris law for the description of crack growth characteristics in visco-elastic media. - The procedure is simple and can be used for practical purposes, especially in the mix design procedure where the quality of various mixes with respect to fatigue and crack resistance has to be evaluated. It is also extremely attractive for specification and pavement evaluation purposes.

7. References [BON 80] Bonnaure, F.P., Gravois, A. and Udron, J. A New Method for Predicting the Fatigue Life Characteristics of Bituminous Mixes, Proceedings of Association of Asphalt Paving Technology (AAPT), Volume 49, pp. 499-529, 1980. [EWA 86] Ewalds, H.L., and Wanhil R.J.H., Fracture Mechanics, Co-publication of Delftse Uitgevers Maatschapij and Edwards Arnold Publishers, London, 1986. [GER 88] Gerritsen, A.H., van Gurp, C.A.P.M, van der Heide, J.P.J., Molenaar, A.A.A. and Pronk, A.C., Prediction and Prevention of Surface Cracking in Asphalt Pavements, Proceedings of the 6th International Conference on the Structural Design of Asphalt Pavements, Ann Arbor, MI, USA; University of Michigan; Volume I, pp. 378-391; Ann Arbor, MI, USA, 1988. [JAC 95] Jacobs, M.M.J. Crack Growth in Asphaltic Mixes, PhD. Thesis, Delft University of Technology, the Netherlands, 1995. [LYT 93] Lytton, R.L., Uzan, J., Fernando, E.M., Roque, R., Hiltunen, D., and Stoffels, S.M., Development and Validation of Performance Prediction Models and Specifications for Asphalt Binders and Paving Mixes, Strategic Highway Research Program, Project A-357, SHRP/NRC, Washington DC, USA, 1993.

Estimation of fatigue characteristics of asphaltic mixes using simple tests

17

[MED 99] Medani, T.O.,A Simplified Procedure for Estimation of the Fatigue and Crack Growth Characteristics of Asphaltic Mixes, M.Sc. Thesis, International Institute for Infrastructural, Hydraulic and Environmental Engineering, Delft, the Netherlands, 1999. [MOL 83] Molenaar, A.A.A., Structural Performance and Design of Flexible Road Constructions and Asphalt Concrete Overlays" Ph.D. Thesis, Delft University of Technology, the Netherlands, 1983. [MOL 00] Molenaar, A.A.A., and Medani, T.O., Rational Testing Methods for Performance Based Specifications Proceedings of the 1st International Conference, World of Asphalt Pavements, Sydney, Australia, pp. 484-500, 2000. [PAR 63] Paris, P.C, and Erdogan F., A Critical Analysis of Crack Propagation Laws, from: Transaction of the ASME, Journal of Basic Engineering, Series D, 85, No.3, 1963. [RIC 68] Rice, J.R., A Path Independent Integral and the Approximate Analysis of Strain Concentrations by Notches and Cracks, Journal of Applied Mechanics, pp. 379-386, 1968. [SAB 95] Sabha, H., Groenendijk, J. and Molenaar, A.A.A., "Estimation of Crack Growth Parameters and Fatigue Characteristics of Asphalt Mixes Using Simple Tests," Delft University of Technology, Road and Railroad Research Laboratory, The Netherlands, 1995. [SCH 73] Schapery, R.A., A Theory of Crack Growth in Visco-Elastic Media, Report MM 2764-73-1, Mechanics and Materials Research Center, Texas A&M University, USA, 1973. [SCH 75A] Schapery, R.A., A Theory of Crack Initiation and Growth in ViscoElastic Media-I Theoretical Development, International Journal of Fracture, Volume 11 No. 1, pp 141-159, February 1975. [SCH 75B] Schapery, R.A., A Theory of Crack Initiation and Growth in ViscoElastic Media-II Approximate Methods of Analysis, International Journal of Fracture, Volume 11 No. 3, pp 369-388, June 1975. [SPD 78] Shell International Petroleum Company Limited, "Shell Pavement Design Manual (SPDM); Asphalt Pavements and Overlays for Road Traffic," London, 1978. [SHR 94] Tayebali, A.A., Deacon, J.A., Coplantz, H.S., Harvey, J.T., and Monismith, C.L., Summary Report on Fatigue Response of Asphalt Aggregate Mixtures, Strategic Highway Research Program, Project A-404, Institute of Transportation Studies, University of California Berkeley, California, June 1994. [DIJ 75] Van Dijk W., Practical Fatigue Characterization of Bituminous Mixes, Proceedings of the Association of asphalt Paving Technologists (AAPT), Volume 44, pp 38-74, 1975.

18

International Journal of Road Materials and Pavement Design, Volume X, No. X/2000

[DIJ 77] Van Dijk W. and Visser W., The Energy Approach to Fatigue for Pavement Design, Proceedings of the Association of asphalt Paving Technologists (AAPT), Volume 46, pp 1-40, 1977.