Material behaviour of mastic asphalt for orthotropic steel deck bridges

Material behaviour of mastic asphalt for orthotropic steel deck bridges Abstract Medani T. O.*, Kolstein M.H., Scarpas A., Bosch, A. and Molenaar A. A. A. Faculty of Civil Engineering and GeoSciences Delft University of Technology P. O. Box 5048, 2600 GA Delft t.medani@citg.tudelft.nl Tel.: 015 2782763 Fax: 015 2783443 Mastic asphalt is normally used for surfacing of orthotropic steel deck bridges. Due to factors such as complicated loading conditions, loss of adhesion between the structural components etc, several problems with the surfacings have been reported including rutting and cracking. Furthermore, it is known that the strain levels in the asphalt layer exhibit extremely high values compared with the strain levels generally encountered in normal asphaltic pavements. In this paper the results of an extensive experimental program carried out at the Road and Railway Research Laboratory (RRRL) of the Delft University of Technology on the mastic asphalt, which was used for resurfacing of the Moerdijk Bridge in June 2000, will be presented. The purpose of the program is to provide an insight into the behaviour of mastic asphalt for orthotropic steel deck bridges. The program included determination of the fatigue resistance and the stiffness master curves. The results of the fatigue testing have shown that mastic asphalt possesses a rather good fatigue resistance when compared with some asphaltic mixes. Furthermore, a very strong dependency of the mix stiffness on the strain level has been observed. Neglecting this phenomenon may result in serious overestimation of the life span of the surfacings. Key Words: Mastic asphalt, orthotropic bridges, fatigue, strain dependency

1 Introduction Modern steel bridge decks consist of a 10-14 mm thick plate stiffened by 6 mm closed longitudinal stiffeners spanning in the direction of the traffic flow between the transverse girders. Usually, the deck plate is surfaced with a 50-70 mm thick surfacing material [Kolstein and Wardenier, 1997]. A typical cross-section is shown in Figure 1. Mastic asphalt Membrane Steel plate 50 10 Stiffener Figure 1. A typical cross section of an orthotropic steel bridge In the Netherlands there are 86 fixed and movable orthotropic steel bridges [Bosch, 2001]. Some of them constitute part of the main highway network, e.g. Moerdijk Bridge, Ewijk Bridge and Van Brienenoord Bridge. Before the 1970 s minor problems with the surfacings of orthotropic steel bridges were observed. However, after that date several problems have been reported, in many countries, including rutting and cracking e.g. Ewijk Bridge in the Netherlands [NPC, 1996] and the Popular-Street Bridge in the USA [Gopalaratnam, 1989]. These problems may be associated with the changes in traffic in terms of number of trucks, heavier wheel loads, introduction of widebase tires etc. As a result of the complicated interaction between the structural components, the different materials involved and the vehicles, strain levels in the order of µm/m have been measured in the surfacings. Such a strain level is quite high when compared with the strain level generally encountered in ordinary road pavements (±200 µm/m). In the Netherlands orthotropic steel bridge decks are normally surfaced with mastic asphalt. Mastic asphalt is an overfilled mix i.e. the voids are overfilled with mastic (bitumen and filler). This results in an extremely small voids content < 2% [Medani, 2001]. The general characteristics of mastic asphalt include: - Due to the overfilled voids and the absence of a steady body, the asphalt is not compacted. - Because of the high percentage of fines, mastic asphalt has a poor skid resistance. To improve the skid resistance fine aggregate (grit) are scattered on the surface. - The stability is not obtained from the mineral body but from a stiff type of bitumen (45/60, 20/30 or a mix from 45/60 and Trinidad-Épurée or modified bitumen) and a large percentage of filler. - The large amount of mastic (thus the very low voids content) render these types of mixes to be extremely impermeable. - Due to this large mortar stiffness the production and processing take place at relative high temperatures (220-240ºC).

- For the production of the mix a normal installation can be used with some special supplies. - For construction the use of a special spreading machine is required. - Because of the high processing temperature isolated transport is commonly used. - The thickness of the layers varies between 20 mm and 70 mm [VBW asfalt, 1996; Whiteoak 1990]. - Because of the special deformation properties, mastic asphalt can not be tested with the Marshall test like the other mixes. 2 Experimental program An extensive experimental program has been designed to characterise the mastic asphalt mix which was used for resurfacing the Moerdijk bridge in June 2000. The testing program has been carried out at the Road and Railway Research Laboratory [RRRL] of the Delft University of Technology. The experimental program included: 1. Determination of the fatigue resistance. 2. Determination of the stiffness master curves. 3. Investigating the dependency of the mix stiffness on the strain level. 2.1 Mix composition The entire testing program has been carried on the mix that was used for resurfacing of the Moerdijk Bridge in June 2000. However, extra testing for investigating the dependency of the mix stiffness on the strain level has been conducted on specimens obtained from slabs casted when surfacing the bridge prototype to be tested at the accelerated testing unit Lintrack. Both the Moerdijk mix and the Lintrack mix have the same mix composition. The mastic asphalt mix consists of stone 2/8 and 2/6 in the ratio 1:1, river sand and fine sand in the ratio 2:3, weak limestone filler and SBS modified bitumen with a pen of 90. The mix composition by mass and volume is shown in Tables 1 and 2 respectively. Table 1. Mix composition of the mastic asphalt mix by mass Component Grain size Percentage by mass Stone 2 mm<ф<8 mm 52.0 Sand 63 µm< Ф <2 mm 26.0 Filler Ф <63 µm 22.0 Bitumen on 100% (by mass) 8.8 Table 2. Mix composition of the mastic asphalt mix by volume Component Percentage by volume Aggregate 63 Filler 17.5 Air Bitumen 1.5 18

2.2 Specimen preparation 1. While resurfacing the Moerdijk Bridge three slabs of dimensions 600x600x70 mm have been casted. From one of these slabs 8 beams of dimensions 450x50x50 mm have been sawn. These beams have been used for the determination of master curves and the fatigue characteristics using the four-point bending test set-up. 2. While surfacing the bridge prototype to be tested at the Lintrack, several slabs have been casted. 46 beams of dimension of 420x50x50 mm have been sawn from four of the slabs. These beams have been used for investigating the dependency of the mix stiffness on the strain level. To avoid deformation of the specimens, they were stored at 13 C in boxes filled with a bed of sand. 3 Experimental results 3.1 Determination of mix stiffness master curves for Moerdijk mix The Universal Testing Machine (UTM) testing equipment was used for the determination of the stiffness master curves i.e. the relationship between the mix stiffness, loading time and temperature. Four specimens were tested, from which the average stiffness values were determined. The test conditions are: - Type of test : displacement controlled - Frequency : 0.5, 1, 2, 5 and 10 Hz - Temperature : 5, 10, 15, 20, 25, 30 o C - Strain amplitude : 80 µ m/m - Wave shape : sine - Measurement of mix stiffness : after100 pulses The stiffness master curves have been determined using the Arrhenius type equation [Francken et al. 1988]. Figure 2 shows the master curves for various temperatures. 00 T=5 deg C Smix [MPa] 0 T=10 deg C T=15 deg C T=20 deg C T=25 deg C T=30 deg C 100 0.00001 0.0001 0.001 0.01 0.1 1 10 100 0 t [s] Figure 2. Master curves for mastic asphalt at different temperatures

3.2 Determination of the fatigue resistance of the Moerdijk mix The UTM machine has been used to conduct the four-point bending beam test. The tests conditions are: - Type of test : displacement controlled - Frequency : 5 Hz - Temperature : 10 o C - Wave shape : sine The frequency and temperature combination has been chosen in such a way to result in a mix stiffness value of approximately 5000 MPa to enable comparison of the results with previous fatigue tests on several candidate mixes for surfacing of orthotropic steel bridges carried out by Kolstein in 1989. Table 3 shows the results of the fatigue tests. Table 3: Results from fatigue testing at 10 o C and 5Hz Strain [µm/m] Load repetitions [-] Initial stiffness [MPa] Dissipated energy [MPa] 375 3600000 4679 3607.006 475 934900 4444 1248.319 600 152630 4083 300.082 750 28140 3443 76.463 925 5430 3820 25.331 1095 4270 3174 24.367 3.2.1 Characterisation of fatigue parameters using the Wöhler approach The phenomenological Wöhler approach is normally used to characterise the fatigue resistance of asphaltic mixes. The Wöhler approach is a relationship between the number of strain applications to failure and the flexural tensile strain, which occurs at the bottom of the test specimen written as: n 1 N f = k1 ε where: N f : number of strain applications to failure ε : flexural tensile strain at the bottom of test specimens [µm/m] k 1, n : factors, depending on the composition and properties of the asphalt mix (1) The values of log k 1 and n, determined by regression analysis, are: log k 1 =23.903, n =6.736 with R 2 =0.987. 3.2.2 Does the Wöhler approach really describe the fatigue resistance of asphaltic mixes? In the Wöhler approach it is assumed that the number of load repetitions is inversely proportional to the strain in a log-log scale i.e. higher strains reduce the life span of pavements. This makes sense, yet in a fatigue test the mix stiffness of the specimens does not remain constant, even if the tests are conducted at the same frequency and temperature. A close look at the results of the fatigue tests shown in Table 3 reveals that the initial mix stiffness has changed with the change of strain (at the same temperature and frequency). In general, it can be said that the stiffness tends to decrease with increasing strain level. The same trend can be observed in fatigue tests carried out e.g. by Kolstein [1989] and at the RRRL.

The authors believe that the fatigue results can be better represented using a Shell type equation [SPDM, 1978] in which the stiffness is included in the fatigue equation. n 1 2 1 1 N f = k1 S ε mix in which: N : number of load repetitions [-] ε : strain [µm/m] : initial mix stiffness [MPa] S mix n (2) Regression analysis has resulted in the following regression constants: log k 1 =33.377, n 1 = 7.439 and n 2 = 2.086 with R 2 =0.99. Figure 3 shows the relationship between the number of load repetitions, the strain level and the mix stiffness for the mastic asphalt mix. Strain (µm/m) 0 100 N=10 2 N=10 4 N=10 6 N=10 8 10 100 0 00 Mix stiffness (MPa) Figure 3. Relationship between the number of load repetitions, the stiffness and the strain for the Moerdijk mix 3.2.3 Comparing the fatigue behaviour of the Moerdijk mix with some candidate mastic asphalt mixes Figure 4 shows the fatigue line of the Moerdijk mix and also the fatigue lines of four candidate mixes for surfacing of steel bridges. The four mixes were REFBL0 (Reference mix with bitumen 45/60 + Trinidad-Épurée), STYL0 (Styrelf 1360), SEAL1 (Sealoflex) and EVAL2 (Bitumen 45/60 + Eva 18-150). It has to be noted that the fatigue tests for these mixes were executed at T=20 o C and f=30 Hz, while the fatigue test for the Moerdijk mix is determined at T=10 o C and f=5 Hz. Nevertheless, these lines are somehow comparable, as the stiffnesses of the different mixes are comparable.

T=20 o C, f=30 Hz 0000 REFB-L0 Load repetitions 000 00 0 STY-L0 SEAL1 EVAL2 Moerdijk mix, 10C 5Hz 10 100 0 Strain [µm/m] Figure 4. Fatigue behaviour of different mastic asphalt mixes From Figure 4 it can be concluded that the fatigue behaviour of the Moerdijk mix is better than that of the other mixes. The figure also shows that the strain range in which the Moerdijk mix is tested is larger than the range in which the other mixes were tested. 3.3 Investigating the dependency of the stiffness on the strain for the Moerdijk mix Two and a half months after the fatigue testing the stiffness master curves were determined again at two different strain levels namely: 80 and 800 µm/m. The same specimens used in the first series of testing were reused again for the determination of the stiffness master curves. Tables 4 and 5 show the stiffness of mastic asphalt at different temperatures and frequencies determined using the four-point bending test set-up. Table 4. Stiffness of mastic asphalt at ε =80 µm/m f [Hz] t [s] 10 o C 20 o C 30 o C 0.5 2 2224 602 262 1 1 2886 757 350 2 0.5 3713 1068 384 5 0.2 5129 1615 490 10 0.1 6264 2099 707 Table 5. Stiffness of mastic asphalt at ε =800 µm/m f [Hz] t [s] 10 o C 20 o C 30 o C 0.5 2 1098 263 83 1 1 1520 374 110 2 0.5 2132 540 159 5 0.2 3490 1095 346 10 0.1 5162 1429 471 The master curves for different temperatures and two strain levels are shown in Figure 5.

0 T=10, strain=80 T=20, strain=80 Smix [MPa] T=30, strain=80 T=10, strain=800 100 T=20, strain=800 T=30, strain=800 10 0.0001 0.001 0.01 0.1 1 10 100 t [s] Figure 5. Master curves after 2.5 months of storage at ε =80 and 800 µm/m From Tables 4 and 5 it can be noted that the mix stiffness does not depend only on the frequency and temperature but also on the strain level. Figure 5 shows that the master curves are shifted to the left with the increase of the strain from 80 to 800 µm/m suggesting that increasing strain tends to reduce the mix stiffness. 3.4 Repeated fatigue tests After the previous series of tests it was decided to repeat the fatigue tests at the same conditions (5 Hz and 10 o C) and on the same specimens. Table 6 shows the results of the repeated fatigue testing. In Figure 6 the results of the first fatigue tests and the repeated fatigue tests are represented using the Wöhler approach. Table 6: Results from repeated fatigue testing at 10 o C and 5Hz Strain [µm/m] Load repetitions [-] Initial stiffness [MPa] Dissipated energy [MPa] 530 1515530 3825 2201.430 600 231970 3910 437.027 700 91040 3541 215.853 740 55070 3620 149.81 800 11390 3759 39.906 920 5770 3610 25.999 0000 first fatigue test 000 repeated fatigue test Load repetitions 00 0 100 0 Strain [µ m/m] Figure 6. Comparison between the first and the repeated fatigue test using the Wöhler approach

From Figure 6 it can be seen that the two fatigue lines are somewhat comparable. Surprisingly, when the strain is less than 700 µm/m, the line obtained from the repeated fatigue test gives higher fatigue life! Although the same specimens were tested before and some of them at a rather high strain level, a considerable recovery of the stiffness has taken place. This may indicates that the material has a rather large healing capacity. 3.5 Determination of the dependency of mix stiffness on strain for the Lintrack mix In previous sections it has been shown that the mix stiffness is dependent on the strain level. However, due to the lack of specimens in the first series of testing, the master curves were determined at only 3 temperatures and 2 strain levels. Therefore it was decided to carry out a more extensive testing program to investigate the dependency of the mix stiffness on the strain. For this additional testing specimens sawn from slabs casted when surfacing the bridge prototype at the Lintrack were used. The test conditions for this additional testing are: - Type of test : displacement controlled - Frequencies : 0.5, 1, 2, 5 and 10 Hz - Temperatures : 5, 12.5, 20, 27.5, 35 and 42.5ºC - Strain amplitudes : 80, 200, 600 and µm/m - Loading wave : sine wave - Stiffness measurement : after 100 pulses The mix stiffness for different strain levels, temperatures and frequencies is shown in Tables 7-10. Table 7: Mix stiffness of mastic asphalt at ε=80 µm/m f [Hz] t [s] 5 o C 12.5 o C 20 o C 27.5 o C 35 o C 42.5 o C 0.5 2 4129 1832 759 367 275 214 1 1 5104 2430 971 484 333 259 2 0.5 6143 3201 1300 573 393 261 5 0.2 7675 4261 1890 850 414 301 10 0.1 8973 5511 2298 1113 572 385 Table 8: Mix stiffness of mastic asphalt at ε=200 µm/m f [Hz] t [s] 5 o C 12.5 o C 20 o C 27.5 o C 35 o C 42.5 o C 0.5 2 3607 1496 539 265 166 121 1 1 4518 1962 728 319 182 136 2 0.5 5613 2701 412 213 143 5 0.2 7283 4006 1673 708 285 158 10 0.1 8730 5187 2228 984 455 302 Table 9: Mix stiffness of mastic asphalt at ε=600 µm/m f [Hz] t [s] 5 o C 12.5 o C 20 o C 27.5 o C 35 o C 42.5 o C 0.5 2 2786 930 367 149 84 60 1 1 3629 1287 500 197 102 62 2 0.5 4706 1786 728 271 125 77 5 0.2 6681 2796 1312 490 295 207 10 0.1 8601 3903 1792 793 456 295

Table 10: Mix stiffness of mastic asphalt at ε= µm/m f [Hz] t [s] 5 o C 12.5 o C 20 o C 27.5 o C 35 o C 42.5 o C 0.5 2 2113 720 279 115 56 42 1 1 2851 1006 399 151 70 51 2 0.5 3945 1418 588 241 129 90 5 0.2 6339 2296 921 469 306-10 0.1 8259 3435 1446 703 437 - Table 10 shows that there are two results missing at a temperature of 42.5 o C. At this high temperature the specimens became so flexible that no measurements were possible at frequencies of 5 and 10 Hz, because in that case the vibration of the specimen caused resonance of the testing machine. The master curves were determined again using the Arrhenius type equation [Francken et al. 1988]. The master curves for different strain levels and some temperatures are shown in figures 8-10. 0 Smix [MPa] eps=80 eps=200 eps=600 eps= 100 10 0.1 1 10 100 0 00 t [s] Figure 8. Stiffness master curve @ T=5 o C 0 Smix [MPa] eps=80 eps=200 eps=600 eps= 100 10 0.0001 0.001 0.01 0.1 1 10 100 t [s] Figure 9. Stiffness master curve @ T=17.5 o C

0 Smix [MPa] eps=80 eps=200 eps=600 eps= 100 10 0.00001 0.0001 0.001 0.01 0.1 1 10 100 t [s] Figure 10. Stiffness master curve @ T=35 o C Figures 8-10 show that for increasing loading time the difference in mix stiffness for the different strain levels is increasing. This means that the strain dependency is increasing with increasing loading times. This behaviour leads to sensibility of the mix to traffic congestion, especially at high temperatures. The maximum difference at high loading times between the mix stiffness at 80 µm/m and µm/m can be a factor 5. 4 Relationship between the mix stiffness, loading time, temperature and strain A relationship between the mix stiffness, loading time, temperature and strain level for the Lintrack mix was found to be best described as: log S mix ( ε, t, T ) = c + d a log ε + b + 1 + e ( log ε ) 1.5 (log t + ft + g + hε ) 0.5 i+ jε 2 (3) where: S mix : mix stiffness [MPa] ε : strain [µm/m] t : loading time [s] T : temperature [ o C] The values of the constants a-j determined by regression analysis are: a : -0.886, b: 3.937, c: 1.273, d: 0.191, f : 0.125, g: -2.297, h: 238.048, i: -1.125, j : -0.014 The relationship is valid within the following conditions: 80 µm/m ε µm/m 0 o C T 45 o C 0.01 s t 1 s It should be emphasised here that the relationship is based merely on the assumption that the beam will still behave elastically at high temperatures. This is because the relationship is

based on measurements obtained from the UTM four-point bending testing machine. The UTM machine is equipped with software, which calculates the stiffness assuming linear elastic behaviour. This assumption might be reasonable for temperatures less than 25 o C. However, for practical purposes, this relationship might be useful. Using equation 3 and by choosing an arbitrary loading time a chart similar to the one shown in Figure 11 can be constructed. 14000 12000 Smix [MPa] 0 8000 6000 4000 T=0 T=5 T=10 T=15 T=20 T=25 T=30 T=35 T=40 T=45 2000 0 0 200 400 600 800 1200 ε [µm/m] Figure 11. Relationship between the mix stiffness, temperature, strain at a loading time of t=0.02 s 5 Effect of neglecting dependency of the mix stiffness on the strain In practice, the master curves are normally determined using low strain levels (typically 80 µm/m). The dependency of the mix stiffness on the strain level is normally neglected. For the design of normal pavements this assumption might be reasonable. However, with the high strain level expected in the surfacings of orthotropic steel bridges ( µm/m) this assumption seems to be quite unreasonable and may be erroneous. For the design of surfacings on orthotropic steel decks theories based on elasticity are normally used in practice (e.g. Metcalf, 1967). With the assumption that theories based on elasticity are applicable for the analysis of such structures, the effect of ignoring the dependency of the mix stiffness on the strain level in the estimation of the life span of the surfacings of orthotropic steel bridges is investigated. In Metcalf s theory the tensile strain in the surfacing is related to the modular ratio (i.e. the ratio of the stiffness of the surfacing material to the stiffness of the steel plate), see Figure 12.

Figure 12. Relationship between the tensile strain in the surfacing and the modular ratio (after Metcalf, 1967) If for instance the stiffness estimated from a master curve, determined at a constant strain of 80 µm/m, a temperature of say 20 o C and a loading time of 0.02 s, is 4700 MPa (see Figure 11), this will result in a modular ratio of 45 and a tensile strain in the surfacing of 700 µm/m. But with such a strain the mix stiffness may drop to 3000 MPa, hence a modular ratio of 70 and a tensile strain of about 900 µm/m. Thus, underestimating the strain by a factor of 1.3, this will result in overestimating the fatigue life span by a factor equal to 1.3 n (where n is the slope of the fatigue line). If we use for instance the n-value determined in section 3, this will yield an overestimation of the fatigue life span by a factor of 5.4. This implies that ignoring the mix stiffness dependency on the strain tends to overestimate the fatigue life span of the surfacing materials. 6 Conclusions The main findings of this research can be summarised at follows: The Wöhler approach should not be used for the characterisation of the fatigue resistance of asphaltic mixes, when it is expected that such mixes will encounter high strains when placed on certain types of structures, e.g. orthotropic steel bridges. In such a case, the use of a Shell type fatigue equation, which includes the stiffness, seems to be more appropriate. The mastic asphalt mix, which was used for resurfacing of the Moerdijk Bridge in June 2000, seems to have a good fatigue resistance when compared with some other candidate surfacing materials. The Moerdijk mix seems to possess a high healing capacity. The mix stiffness seems not to be dependent only on the frequency and temperature but also on the strain level. An equation, which relates the mix stiffness to the loading time, temperature and strain, has been developed. Neglecting the dependency of the mix stiffness on the strain level can lead to overestimation of the fatigue life span of the surfacing materials of orthotropic steel bridges.

7 References Bosch, A., Material Characterisation of Mastic Asphalt Surfacings on Orthotropic Steel Bridges, M.Sc. Thesis, Delft University of Technology, the Netherlands, 2001. Francken, L. and C. Clauwaert, Characterisation and structural assessment of bound materials for flexible road structures, Proceedings 6th international conference on the structural design of asphalt pavements, Ann Arbor, MI, USA, 1988, p. 130-144. Gopalaratnam, V.S., Baldwin, J.W., Hartnagel, B.A. and Rigdon, R.A., Evaluation of Wearing Surface Systems for Orthotropic Steel Plate Bridge Decks, Report 89-2, Missouri Highway Transportation Dept, University of Missouri, USA, 1989. Kolstein, M.H. and J.H. Dijkink, Behaviour of modified bituminous surfacings on orthotropic steel bridge decks. Madrid: 4th Eurobitume symposium, 1989. Kolstein, M.H. and Wardenier, J., Stress Reduction due to Surfacing on Orthotropic Steel Decks, Proceedings ISAB Workshop, Evaluation of Existing Steel and Composite Bridges, Laussanne, 1997. Medani, T.O., Asphalt Surfacing Applied to Orthotropic Steel Bridge Decks, Report 7-01- 127-1, Road and Rialway Res. Lab., Faculty of Civil Engineering, Delft University of Technology, the Netherlands, 2001. Metcalf, C.T., Flexural Tests of paving Materials for Orthotropic Steel Plate Bridges, Highway Research Record No. 155, Washington, D.C, 1967. NPC, Research into Mastic Asphalt Mix of the Ewijk Bridge (in Dutch), Utrecht, the Netherlands, 1996. SPDM, Shell Pavement Design Manual. London: Shell International Petroleum Company Ltd., 1978. VBW asfalt, Asphalt in Road Construction (in Dutch), 9e herziene druk. Breukelen: VBW asfalt, 1996. Whiteoak C.D., The Shell bitumen handbook. Chertsey. Surrey: Shell bitumen U.K., 1990.