Keywords

1 Introduction

During a pavement service life, it is well known that the bituminous binder hardens and becomes more brittle, resulting in an increased risk for different types of crack formation. The bitumen hardening can be simulated in laboratory tests, such as the rolling thin film oven test (RTFOT) or the pressure aging vessel (PAV).

In literature, differences between binder properties due to field aging and due to accelerated laboratory aging tests have been reported [1,2,3,4,5,6,7]. For example, Daniel et al. [1] and Reinke et al. [2], observed that field aging, specifically for recovered top layer material, appeared to be more age hardened as would be predicted by standard protocols. In van Lent et al. [5] the authors report that because of higher temperatures and higher pressures in for example the pressure aging vessel (PAV), and due to the absence of other pavement constituent such as filler, sand and aggregates, unwanted ageing side-effects may take place in the binder. These authors propose to perform laboratory binder ageing directly in an asphalt concrete mixture. Among other recommendations, Hagos proposed the use of a wheatero meter [3]. Li et al. [6] observed minimal differences for an unmodified binder between lab and field aging. But, for two PmBs, the polymer effect was still present after PAV aging, while it was less clear in the recovered binders.

In the study presented here, the purpose is to show initial results on the comparison of field aging versus the long-term PAV aging procedure. The findings are part of a larger study, and in ref. [8] general objectives and findings have been reported. In this paper some representative field sites are presented.

2 Experimental

2.1 Materials

Details of the four field sites, selected for this paper are presented in Table 1, including the service life, the binder type, the binder content and the void content. In the last column the number of slices obtained from each wearing course is indicated. Before cutting the cores in layers, the layer thicknesses, were marked on the cores. Typically as follows; layer 1: 0–0.5 cm, layer 2: 0.5–1 cm, layer 3 1–2 cm, further layers every cm. Afterwards the cores were sliced on these markings, and for each layer the binder was recovered separately. Recovery was conducted according to EN 12697 using trichloroethylene as the solvent. Before the recovery the bulk density of the cores was measured by gamma rays according to EN 12697-7.

Table 1 Overview of test sites
Full size table

2.2 Test Methods

DSR measurements were conducted using Anton Paar Rheometers; frequency temperature sweeps from 0.01 to 10 Hz were recorded with the 25, 8 and 4 mm plate-plate geometry: 25 mm plate test were recorded at 1 mm gap, 1% strain and from 50–90 °C, in steps of 10 °C. 8 mm plate tests at 2 mm gap, 0.05% and from 0–50 °C, 4 mm plate tests at 1.75 mm gap, 0.02% strain and from 10 to −24 °C in steps of 6 °C (except between 10 and 0 °C). A detailed description can be found in [9]. Different rheological parameters were calculated, according to procedures explained in [8].

FTIR measurements were conducted with a Nicolet IS 10 with a diamond cell and an attenuated total reflection setup. The carbonyl absorption was used as an aging parameter, the procedure how this was integrated is also explained in [8].

For the fractionation, the asphaltenes were first separated based on IP 143, afterwards the maltenes were further divided based on IP 469.

Aging tests were conducted according to EN 14769. For the recovered binders smaller containers were used, adapting the amount of binder as is recommended in EN 14769.

3 Results

3.1 Evaluation of the Aging in the Wearing Course

In Fig. 1, delta Tc and G-R levels are presented for each of the slices of the four test sections. The orange lines represent limiting levels, as proposed in literature. For the French section, there is almost no difference between the three slices, while for the other sections, the bottom slices are clearly less aged (less negative Delta Tc and a lower G-R) as compared to the top slices. This can be related to the void content, Sect. 1 is a very open mix with an average void content of 12.3%, and all the layers are about equally aged, the Belgian and UK section have intermediate voids of 6.1 and 8% respectively, and especially for the UK section a clear gradation is seen as the slices correlate to deeper sections. The Croatian section is the densest mix with a void content of 2.5%, and for this section only the thin top layer is aged. This section also shows the largest difference between top and bottom parts. A similar variability was observed for the other rheological parameters related to aging.

In Fig. 2, some chemically related parameters are presented, such as the carbonyl index and the SARA fractionation data. The carbonyl index was calculated for each recovered binder, [8] while SARA fractions were determined for the top and bottom slice only, for the three sections for which the top and bottom slices were different.

The carbonyl index follows a trend that is very similar as observed for the mechanical properties. For the SARA fractions the changes can be summarized as follows, the saturates remain about constant, the aromats clearly decrease from bottom to top, the polar I (resins) fraction increases for two of sites, but decreases in one case, while the asphaltene fraction clearly increases for all the sections. These changes correspond to what is observed after laboratory aging tests. As the saturate fraction remains about constant, it indicates that a further volatilization during service life is unlikely.

Fig. 1
figure 1

Data from the recovered binders: a Delta Tc levels, b G–R, c SARA fractions, d the carbonyl index

Full size image

3.2 Comparison to Laboratory Aging Tests

Initially, the properties of the recovered binders, from all the slices, were compared to the properties before and after laboratory aging obtained on a reference set of binders. The reference set, all unmodified binders, was collected in the market between 2012–2014. The comparison for the G-R parameter is illustrated in Fig. 2a. The reference binders before aging are collected in the blue circle in Fig. 2a, while after RTFOT + PAV they have moved into the larger orange shape. The recovered binders are also plotted; For the unmodified sections, all the slices from the French section, and the top slice of the Croatian section are aged beyond the levels observed in the reference set. For the French section, this could possibly be influenced by the presence of a small amount of RA binder, for the Croatian top slice it is related to field aging. For both PmB sections, the Belgian section is borderline if compared to the reference section, while the UK section is clearly outside of what is observed in the reference section. This is most likely related to the polymer modification.

As the comparison to the reference set is still relative, a further aging of the bottom slices was conducted for the three sections, displaying a difference between top and bottom. The result of the PAV aging of the bottom slice is shown in Fig. 2b, black squares. These are coded as follows; the country code & slice number followed by PAV. For the unmodified binder (HR), and one of the PmBs (BE), the aging level for the field aged top sample is clearly larger than the PAV aged binder, but the slope of the curve, when moving from the bottom to the PAV aged binder is almost similar as for the field aging. For the second PmB section (UK) the difference between lab aging and field aging is larger. This is most likely related to the polymer modification. In the UK section the polymer in the bottom slice is more intact as compared to the Belgian section, since it was from a deeper slice, which may explain the larger effect.

Fig. 2
figure 2

a Comparison of G-R levels of the recovered binders versus the reference set. b The effect of aging of the bottom slice in a standard PAV test on the G-R levels

Full size image

4 Conclusions

In conclusion, the trends observed in the properties of field aged binders are similar to the changes seen or expected to happen in accelerated aging tests. In the field, in some cases, binders seem to age more as predicted by the laboratory long term PAV under standard conditions. For one of the unmodified sites, this could be related to the presence of a small amount of RA, for the other unmodified binder, the most logical explanation is that field aging just has continued to this level over the long service life. Another option, as proposed in literature, could be related to the aggregate or filler material which may accelerate field aging, and which is absent in the PAV aging. Small differences in the relative magnitude of the changes (phase angle versus stiffness shifts) have been observed in the G-R black diagram. For two of the binders these changes are very small, one unmodified and one PmB, while for the second PmB the differences are larger. The aging depth in the wearing course relates to the void content of the pavement: For dense pavements only a thin slice at the very top of the layer is aged, for intermediate void contents a gradual change in the degree of aging through the wearing course is observed, while for the open pavement the whole layer is equally aged.

Some uncertainties remain: for example, regarding performance properties, a more aged binder deteriorates the properties. But, if the aged top part in the wearing course is very thin, 0.5 cm, it is not clear how important this thickness is. Maybe a minimum thickness, of the aged layer, needs to be present before being able to generate damage. And this minimum could depend on the mix type or also failure type, cracking or raveling. Regarding the suitability of the PAV test to simulate field aging, at this moment the data presented is not enough to conclude that the standard PAV test is not adequate. More data should be collected, before drawing this conclusion.