PRACTICAL TEST METHODS FOR MEASURING THE ZERO SHEAR VISCOSITY OF BITUMINOUS BINDERS

Transcription

1 124 PRACTICAL TEST METHODS FOR MEASURING THE ZERO SHEAR VISCOSITY OF BITUMINOUS BINDERS J. De Visscher and A. Vanelstaete Belgian Road Research Centre, Belgium Abstract When the zero shear viscosity (ZSV) is derived from a creep test, the creep time should be sufficiently long to obtain a steady state with a constant shear rate. The creep test works well for most conventional binders. Unfortunately, for polymer modified binders (PmBs), the long waiting times required to obtain a steady state are not always acceptable from a practical point of view and there are problems concerning the repeatability. This paper compares the steady state creep test to alternative test methods for the determination of the ZSV. One method is a short time creep/recovery test, with a short creep period and a longer recovery period, where the ZSV is derived from the recovery part. The other method is an oscillation test at low frequencies, where the complex viscosity is extrapolated to zero frequency to obtain the ZSV. Measurements are presented on a pure 70/100 binder and on a highly SBS-modified binder. While all methods give comparable and repeatable results for the case of the pure binder, complications occur for the highly modified binder. The oscillation method seems to be the most promising method. 1. Introduction The rheology of the binder is a major factor influencing the permanent deformation of asphalt pavement. In the battle against rutting, it is essential to develop a reliable test method for measuring the relevant rheological properties in order to make an appropriate selection of the binder. The Superpave binder test methods use G/sin, measured at =10 rad/sec, as a performance indicator for rutting [1]. In the case of pure binders, the correlation between this indicator and results from rutting tests on asphalt mixes (e.g. wheel tracking tests) is good. For PmBs on the other hand, the Superpave indicator generally underestimates the resistance to rutting [2, 4]. In recent publications on Superpave tests for Pmbs [5, 6], this phenomenon was investigated and explained by the effect of delayed elasticity. A repeated creep test was then developed to derive a more appropriate rutting parameter for PmBs. In Europe, attention was focussed on the parameter called the zero shear viscosity (ZSV) [2, 3]. In a long time creep test, the effect of delayed elasticity decreases with time and after a sufficiently long period, the rheological behaviour is dominated by viscous flow. Viscous flow is characterized by the Newtonian viscosity, also called the zero shear viscosity (ZSV), since it is to be measured at very small shear rates. A problem is that, in the case of PmBs, the time scale of delayed elastic

2 Performance Testing and Evaluation of Bituminous Materials 125 deformation is so large that it can take hours or days before reaching a steady state of purely viscous flow. Highly modified binders may even never reach a steady state [7]. In this paper, alternative types of tests for identifying the ZSV are compared, with the aim of evaluating experimental time and effort while obtaining repeatable results. 2. Identification of the ZSV 2.1 Creep tests A bituminous material subjected to a constant stress exhibits a compliance curve, which is a superposition of contributions from instantaneous elastic deformation, delayed elastic deformation and viscous flow (fig. 1). When the period of the creep test is long enough, the deformation attains a steady state governed by viscous flow. The slope of the compliance curve is then inversely proportional to the ZSV: 1 ZSV ( dj c dt) for t (1) where J c is the creep compliance. For some materials, the time required to attain steady state behaviour may be too long from a practical viewpoint. Also, long creep periods involve high strains, even when the applied stress is small. This may induce microstructural changes in the material, leading to problems with repeatability. deformation delayed elastic deformation viscous flow total deformation time Fig. 1: Deformation components in a creep test 2.2 Recovery tests At the time the stress is relieved, instantaneous elastic deformation immediately disappears and delayed elastic deformation slowly returns to zero. When the recovery compliance has reached a constant value, this value corresponds to the contribution of unrecoverable viscous flow. The ZSV is thus obtained as follows: ZSV t J (t) for t (2) creep r where t creep is the period of the creep test and J r is the recovery compliance. The recovery test method can also be used after a short period creep test, since it is not necessary that a steady state was attained in the preceding creep test. The main motivation for using this short time creep/recovery test is that the strain level is limited. 2.3 Oscillation tests When the frequency tends to zero, the complex viscosity approaches a value equal to the ZSV: 1 ZSV ( ) ( J ( ) ) for 0 (3)

3 126 where is the complex viscosity and J the complex compliance. The 4-parameter Cross model can be used to extrapolate measurements of the complex viscosity to zero frequency [8]: 0 ( ) m 1 ( K ) where the parameter equals the ZSV. 0 It may be argued that an oscillation test is not suitable to estimate the susceptibility to permanent deformation, since the reversal of deformations does not allow separating the contributions from reversible and non reversible (permanent) deformation to the total dissipated energy. This argument is correct and it explains the deficiency of the SHRP parameter G/sin (at =10 rad/sec). However, when the oscillation frequency tends to zero, the contribution of reversible deformation also tends to zero and the ZSV does become a liable candidate indicator for permanent deformation. 3. Experimental work Three types of tests were performed: Long time creep/recovery tests, where ZSV was derived from both creep and recovery compliance (eq. (1) and (2)); Short time creep/recovery tests, where ZSV was derived from the recovery compliance (eq. (2)); Strain controlled oscillation tests in the range from 10-3 Hz to 20 Hz, where ZSV was obtained by fitting the data with the Cross model (eq. (4)). All tests were made with the Bohlin CVO120 Rheometer with parallel plate geometry ( 25 mm). The test temperature was 40 C, but this is not necessarily the optimum test temperature for all binders. Before testing, a strain sweep at =10 rad/sec was made to verify the range of linearity. 3.1 Results for pure binder (B 70/100) Table 1 shows the results from long time creep/recovery tests on a pure 70/100 binder sample B0 (creep: 1h, recovery: 1h, shear stress: 500 Pa). The test was repeated 5 times. The last column shows the ZSV derived from the slope of the creep compliance after 15 minutes. From the 4 th run, the results are stable. The difference between the results from creep and recovery is then only 0.1 %. There is also less than 0.2 % difference between the creep test results at 15 min and 1 hour. The test period for this binder can thus be reduced to 15 min. To study the repeatability, 4 other samples (B1, B3, B5 and B6) of the same binder were subjected to the test (table 2). The creep and recovery periods were reduced to 15 minutes, as justified by the preceding test on sample B0. From the 2 nd run, the variations from run to run were less than 1 %. In the 4 th run, an average value of 11.9 kpas was obtained with a standard deviation of 1.5 kpas for both creep and recovery. (4)

4 Performance Testing and Evaluation of Bituminous Materials 127 Table 1: Long time creep/recovery on pure 70/100 binder sample B0 (creep: 1h, recovery: 1h) run t=3600 sec t=3600 sec t=900 sec Table 2: Long time creep/recovery on pure 70/100 binder (creep: 15 min, recovery: 15 min) run t=900 sec t=900 sec B0 B1 B3 B5 B6 Av. St.dev. B0 B1 B3 B5 B6 Av. St.dev : except for sample B0, t=3600 sec The short time creep/recovery test was performed with a creep period of 10 sec and a recovery period of 100 sec. This recovery period was found to be sufficient to obtain a steady value for the recovery compliance. The results are summarized in table 3. A creep period of 10 sec is too short to attain a steady state in creep. This explains why the creep results are systematically smaller. The average of the results from the recovery part in the 4 th run is 12.0 kpas, which is close to the average of the long time creep/recovery tests. The variability between samples is probably due to sample preparation and variability of sample size. If more samples were tested, the standard deviations of both types of test are expected to converge. Table 3: Short time creep/recovery on pure 70/100 binder (creep: 10 sec, recovery: 100 sec) run t=10 sec t=100 sec B00 B1 B2 B4 B7 Av. St.dev. B00 B1 B2 B4 B7 Av. St.dev The oscillation test was first applied to samples B10 and B11 in 2 runs (table 4). The results are in the same range as the average results from the creep/recovery tests. Since there was an increase from run 1 to 2, the test was repeated 5 times on another sample B12, also shown in table 4. Figure 2 shows the experimental data and the fit for the last run. An increase is again observed from run 1 to 2, but from then on the value remains relatively stable. However, the value is small. It is possible that sample B12 was an outlier, but a statistically larger number of tests would be necessary to confirm this.

5 128 Table 4: Oscillation tests on pure binder 70/100 (10-3 to 20 Hz) run B10 B11 B / / / / / / 7.5 Complex viscosity (Pas) 8.E+03 6.E+03 4.E+03 2.E+03 0.E+00 1.E-03 1.E-02 1.E-01 1.E+00 1.E+01 1.E+02 freq (Hz) Fig. 2: Oscillation test on pure binder 70/100 sample B PmB with high polymer content (7.5% SBS) Table 5 shows the results from long time creep/recovery tests (creep: 4h, recovery: 4h, shear stress: 50 Pa) in 5 subsequent runs. Although the results in creep and recovery are the same, the repeatability of the test on the same sample is not satisfactory. The ZSV continues to increase from run to run. A similar increase for highly modified binders was reported by other researchers [7]. It was explained by the phase morphology of the binder: when the polymer phase becomes continuous, which occurs for highly modified binders, subsequent stretching of the sample modifies the morphology of the phases. Table 5: Long time creep/recovery on highly modified binder (creep: 4h, recovery: 4h) run t=4hr t=4hr t=1hr Short time creep/recovery tests were performed with a creep period of 10 sec and recovery periods of 100, 500 and 900 sec respectively. These tests showed that the recovery compliance had not attained a constant value, as seen in figure 3 for a recovery period of 900 sec. Table 6 shows the ZSV derived from the tests in figure 3. fit exp

6 Performance Testing and Evaluation of Bituminous Materials 129 compliance (1/Pa) 1.2E E E-04 run 1 run 2 run 3 run 4 run 5 3.0E time (sec) Fig. 3: Creep (10 sec) / recovery (900 sec) curves for highly SBS-modified binder Table 6: Short time creep/recovery on highly SBS-modified binder (creep: 10 sec, recovery: 900 sec) run t=10 sec t=900 sec It is interesting to note that, although the ZSV derived from the creep curve at 10 sec is an underestimation, the result is repeatable on the same sample. The strains are indeed much smaller than for the long time creep test. The ZSV derived from recovery decreases from run to run, but it seems to attain a stable value from run 4. More runs should confirm this. The oscillation test was performed on 2 samples, B20 and B21, in the range from 10-3 to 20 Hz (5 runs). Table 7 shows the results. Also included are the results obtained when the fit is confined to the data in the range from 10-2 to 20 Hz. The results are close. This means that the lower frequency limit could be set to 10-2 instead of 10-3 Hz, leading to a considerable time saving. Table 7: Oscillation tests on highly SBS-modified binder B20 B21 run 10-3 Hz to 20 Hz 10-2 Hz to 20 Hz 10-3 Hz to 20 Hz 10-2 Hz to 20 Hz Discussion Comparing rheological results from different test methods is not easy, since the variability in preparing the small binder samples is relatively high. A large amount of tests is necessary to obtain statistically relevant data. The data presented in this paper is rather limited, since the research is still in an exploratory stage. However, some interesting observations can already be made to give a direction to future research. Long time creep tests present a problem of repeatability, when applied to a highly modified binder. A possible explanation has been proposed in terms of the high strains and the morphology of the binder which is affected by these high strains.

7 130 Short time creep/recovery tests are an interesting alternative, since the strain level is smaller. Still then, the recovery period has to be long to attain a constant recovery compliance. For the highly SBS-modified binder, a recovery period of 900 sec was not enough. When applying this method, a compromise should be made between acceptable testing time and a sufficient level of accuracy. With the oscillation test, stresses and strains are constantly reversed, so that overstretching of the sample is avoided. The results on the highly SBS-modified binder show that it is not necessary to measure at extremely small frequencies. The Cross model is capable of extrapolating the data to zero frequency. It should also be noted that data measured at very low frequencies is less reliable, since the stress levels are too small in a strain controlled test. The oscillation test is considered as an interesting test to determine the ZSV, even for highly modified PmBs. 5. Conclusions Three types of tests for identifying the ZSV were compared, a long time creep/recovery test, a short time creep/recovery test and an oscillation test. The tests were made on a pure 70/100 binder and a highly modified binder (7.5 % SBS). The amount of test results is still limited and other binder types should be considered in a future study. The present conclusions should therefore be considered as preliminary. For the pure binder, the tests were repeatable and the results agreed sufficiently well. The tests on the highly modified binder on the other hand revealed a problem of repeatability for the long time creep tests. The oscillation method did not reveal this problem. Oscillation tests also have the advantage that the experimental time and effort are acceptable and that a large amount of interlaboratory work has already been done in the field of DSR testing. Acknowledgements The authors thank Mr. P. Peaureaux for his help with the experimental work and the Ministère des Affaires Economiques for the financial support in this research project. References 1.Superior Performing Asphalt Pavements: The product of the SHRP Asphalt Research Program, SHRP-A- 410, National Research Council, Phillips, M. and Robertus, C. Binder rheology and asphaltic pavement permanent deformation; the zeroshear viscosity, Proc. Eurasphalt & Eurobitume Congress 1996, paper Sybilski D., Validation of empirical tests for polymer-modified bitumens, Proc. Eurobitume Workshop 99, Paper n Gershkoff, D. et al., The influence of binder properties on the wheel-tracking rate of hot rolled asphalt, Proc. Eurasphalt & Eurobitume Congress 1996, paper NCHRP report 459, Characterization of Modified Asphalt Binders in Superpave Mix Design, National Research Council, Bahia H.U. et al., Development of binder specification parameters based on characterization of damage behavior, Proc. of 2001 meeting of AAPT, pp Desmazes, C. et al., A protocol for reliable measurement of zero-shear viscosity in order to evaluate the anti-rutting performance of binders, Proc. Eurasphalt & Eurobitume Congress 2000, Book 1 pp Cross, M., Rheology of non-newtonian fluids: a new flow equation for pseudoplastic systems, J. of Colloid Science 20 (1965)