Investigating self healing behaviour of pure bitumen using Dynamic Shear Rheometer

Fuel 90 (2011) 2710–2720

Contents lists available at ScienceDirect

Fuel journal homepage: www.elsevier.com/locate/fuel

Investigating self healing behaviour of pure bitumen using Dynamic Shear Rheometer J. Qiu a,b,⇑, M.F.C. van de Ven a, S.P. Wu b, J.Y. Yu b, A.A.A. Molenaar a a b

Road and Railway Engineering, Civil Engineering and Geosciences, Delft University of Technology, 2600 GA Delft, The Netherlands Key Laboratory of Silicate Materials Science, Engineering of Ministry of Education, Wuhan University of Technology, Wuhan 430070, China

a r t i c l e

i n f o

Article history: Received 24 December 2010 Received in revised form 8 March 2011 Accepted 10 March 2011 Available online 24 March 2011 Keywords: Self healing Bitumen Dynamic Shear Rheometer Crack closure Normal force

a b s t r a c t Self healing is known as a built-in property of bitumen, which can help bitumen and asphalt concrete to recover its strength after damage. Healing can also extend the service life of asphalt pavements. A twopiece healing (TPH) test was developed to investigate the self healing behaviour of pure bitumen using the Dynamic Shear Rheometer (DSR). During the TPH test, a healing process was directly simulated by pressing the two pieces of bitumen together under a parallel-plate system. Two phases can be distinguished from the TPH healing test, the initial healing phase due to gap closure, and the time dependent healing phase. The results are summarised as following. (1) Initial healing phase: The initial healing phase has a three-stage complex modulus increase with the closure of the gap thickness. A rapid increase of the complex modulus is observed in the second stage of the initial healing curve due to a 0.5 mm gap reduction. (2) Time dependent healing phase: The time dependent healing results show a distinctive difference between the gap constant control mode and the normal force constant control mode. The normal force control healing can be decomposed into the time dependent healing during a gap constant mode and additional healing by compression. It was indicated that the compressive normal force strongly promotes healing development. (3) It was also demonstrated that many factors can influence the complex modulus measurement results by using the DSR parallel-plate geometry. Ó 2011 Elsevier Ltd. All rights reserved.

1. Introduction The so-called self healing capability of asphalt concrete has always been an interesting topic since it was discovered some 40 years ago [1–3]. The significance of the self healing capability is that an asphalt concrete could repair itself under certain conditions such as hot summers or rest periods and extend the service life [4]. Hence, the self healing capability is defined as the recovery of the mechanical properties like strength, stiffness as well as an increase of the number of load repetitions to failure. In the past, investigations have been carried out to measure the healing phenomenon on the scale of asphalt concrete [3,5–7]. It is known that the healing phenomenon of asphalt concrete is highly dependent on its bituminous binders. Hence, knowledge on the healing property of the pure bitumen at smaller scale such as meso- or micro-scale is important to understand its contribution to the self healing capability of asphalt concrete. The Dynamic Shear Rheometer (DSR) was used in this paper to investigate the self healing capability of pure bitumen with a ⇑ Corresponding author at: Road and Railway Engineering, Faculty of Civil Engineering and Geosciences, Delft University of Technology, 2628 CN Delft, The Netherlands. Tel.: +31 (0) 15 27 82763; fax: +31 (0) 15 27 83443. E-mail addresses: [email protected], [email protected] (J. Qiu). 0016-2361/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.fuel.2011.03.016

special two-piece bitumen healing setup. The healing process was investigated separately with initial healing and healing in time. In addition, influence of measuring factors on the experimental results was also discussed.

2. Literature studies on self healing capabilities of bituminous materials using the Dynamic Shear Rheometer The DSR was employed to investigate the rheological properties of bituminous materials at micro- and meso-scales including the dynamic and fatigue properties of bituminous binders, mastics (bitumen/filler systems) and mortars (bitumen/filler/fine sand systems) [8–12]. The DSR was also used to investigate the self healing capacity of bituminous binders. Table 1 gives a overview of the self healing investigations. Three types of self healing tests were developed: Type 1, the discontinuous fatigue test (DFT) using a cyclic strain or stress controlled loading with different rest/load period ratios, in which the healing behaviour was investigated by the extension in fatigue life due to the rest periods [13–18]; Type 2, the fatiguehealing-re-fatigue test (FHRT) using a long healing duration between two continuous fatigue tests, in which the healing behaviour was quantified by means of modulus recovery and/or

2711

J. Qiu et al. / Fuel 90 (2011) 2710–2720 Table 1 List of healing tests using the DSR. Test types Samples

Tempera- Controls Systems tures (°C)

Frequency Stress / (Hz) strain levels

DFT

Bitumen + sand

25

Strain

Column

10

DFT DFT DFT

15 15, 25 15

Stress Stress Stress

Parallel plate Parallel plate (8 mm) Column

FHRT FHRT FHRT

Bitumen Bitumen Bitumen + filler + sand maximum particle size of 0.5 mm Bitumen Bitumen + filler Bitumen

10 10 20

Strain Strain Strain

FHRT

Bitumen

25, 20

FHRT

Bitumen

FHRT FHRT

Healing procedures

Refs. [13,14]

25 10 10

0.35% and 0.9% Two-minute rest periods were applied 10 times during the tests 184–237 kPa 20 s loading/10 s rest–20 s loading /400 s rest 60–230 kPa 1 s loading/0 s rest–1 s loading/6 s rest 0.025–0.1 N m 3 s loading/9 s rest

Parallel plate Parallel plate Parallel plate

41 40 1.6

1.6% 0.3% 20%

[19] [20] [21]

Stress

Parallel plate (8 mm)

10

400 kPa

15

Strain

Parallel plate

25

1.6%

Bitumen Bitumen

5–9 15

Strain Strain

Parallel plate (8 mm) Parallel plate (8 mm)

1.59 25

1% 1.8%

TPH

Bitumen

25

Strain

Parallel plate (25 mm) –

–

TPH

Bitumen

20

Strain

Parallel plate (25 mm) –

–

extension of cycles (load repetitions) to failure in the re-fatigue test [19–24]. Type 3, the intrinsic two-piece healing test (TPH) [25–28]; the details of this test will be explained later. For an indepth discussion of these approaches the reader is referred to the literature review made by Qiu [29]. The DFT tests and the FHR tests indicated the evidence of the self healing behaviour of bituminous binders. The extension of loading cycles to failure which can also be named as the fatigue strength, together with the modulus recovery are the most commonly used indexes to quantify the self healing capabilities. However, a great difference between modulus recovery and fatigue strength recovery was discovered. It was shown that the recovery of stiffness was not directly related to the recovery of fatigue strength. From the DFT tests and the FHR tests one can conclude that the self healing phenomenon exists and is complex. For directly simulating the crack healing process, the first intrinsic TPH test was developed by Little and his colleagues [25–27]. Two pieces of bitumen were placed on the upper and the bottom plates of the DSR. Then the DSR pressed the two pieces of bitumen together to simulate the crack healing process. The change of the shear modulus was measured with a strain level of 0.001% and assisted by a constant compressive force of 0.4 N. The results indicated that the initial healing values obtained by means of gap closure showed a good agreement with the surface free energy values of the five different types of bitumen tested. A model was proposed to describe the healing of bituminous binder based on a convolution of wetting and intrinsic healing processes [25–27]. This model was in line with the multi-step healing model proposed by Phillips [19]: Step (1) flow, Step (2) wetting and Step (3) inter-diffusion. However, Step (1) flow was replaced by an external pressure to ensure the closure of the simulated crack. Qiu also applied a similar test setup for preliminary investigation of the self healing property of different binders [28]. The change of the shear modulus was monitored at a constant gap thickness after the two pieces of bitumen were pressed together. A similar trend was shown from the results when compared with the data reported by Little [25–27]. The friction between two pieces of bitumen samples was observed as an influencing factor

Strain level of 0.02% for 0–5000 s Strain level of 0.003% for 2 h After 5000 cycles, specimen rested for periods of 0.5, 1, 3, and 12 h, then test again Complex viscosity |g⁄| was equal to 50% of the |g⁄| at cycle 10; test was stopped for a rest period ranging from 0 to 48 h, then test started again Four hours for the first rest; 17 h for the second rest Strain level of 0.01% for 2 h Strain level of 0.05% for 500 rest (500 loading/ 500 rest cycles) A constant normal force of 0.4 N, strain amplitude of 0.001% and at 10 rad/s for 1 h A constant gap thickness of 7.8 mm, strain level of 0.00625% at 10 Hz for 2 h

[15] [16,17] [18]

[22]

[15] [23] [24] [25–27] [28]

for the TPH measurement of a hard bitumen with the penetration grade less than 10. By mimicking the crack closure using the DSR, the TPH test is believed to provide a direct relation to the fundamental healing process. Hence, The TPH test is used in this paper to investigate the real self healing behaviour of pure bitumen. 3. Experimental 3.1. Materials A standard 70/100 penetration grade Kuwait Petroleum bitumen with a penetration of 93 (0.1 mm) at 25 °C, and a softening point of 45 °C was used in this study. Pure bitumen specimens were made into a disk shape with a diameter of 25 mm, and the thickness varied between 6 and 6.5 mm for the one-piece sample, and 3–3.5 mm for the two-piece sample. 3.2. DSR and normal force sensor As it is shown in Fig. 1, a Dynamic Shear Rheometer AR2000ex was used applying a 25 mm parallel-plate system [30]. A normal force sensor was located under the bottom plate. The normal force data was investigated intensively in this paper for the following functions: (a) monitoring the normal force (vertical force) development during direct tension and/or compression with a constant change of the gap thickness; (b) measuring the generated normal force during a time sweep test at a constant gap thickness; (c) controlling the constant normal force during a time sweep test by continuously adjusting the gap thickness. 3.3. Research plan Fig. 2 shows the two phases of the TPH test: Phase 1, the initial healing phase due to gap closure; Phase 2, the time dependent healing. In order to analyze the real healing signal and the influencing factors, the two phases were investigated separately in this

2712

J. Qiu et al. / Fuel 90 (2011) 2710–2720

For the two-piece sample, three stages of the development of the complex modulus can be observed.

Fig. 1. Illustration of Dynamic Shear Rheometer AR2000ex [30].

 First stage, when the gap thickness between parallel plates decreases from 7000 lm to 6500 lm, the two pieces of bitumen samples, which were attached to the top plate and the bottom plate, respectively, were visually in touch with each other. This was observed by the web camera inside of the temperature cabinet. However, the measured complex modulus was less than 1 MPa, and the decrease of the gap thickness did not cause much improvement of the complex modulus. The reason could be that the two pieces of bitumen were not really in touch with each other yet, although visual inspection indicated that they seemed to be pressed together.  Second stage, when the gap thickness decreases from 6500 lm to around 6000 lm, the complex modulus increases enormously, which may be due to the contact between the two bitumen pieces.  Third stage, when the gap thickness continuously decreases from 6000 lm to lower, the behaviour of the two-piece sample is almost the same as that of the one-piece sample. Here, the two-piece sample is believed to be fully healed and behaves the same as the one-piece sample. Considering the uniqueness of the three-stage initial healing curve, the increase of the measured complex modulus can be calculated as a function of the gap thickness decrease. The second stage of the initial healing curve can be used as an indicator of the initial value at the time dependent healing data, which will be discussed later.

Healing percentage by Modulus recovered

100%

4.2. Influencing factors on initial healing Phase 2: Time dependent healing

0 Healing time Fig. 2. Illustration of the TPH test.

research. Table 2 shows the main test program of the TPH test. Apart from the healing measurement of the two-piece sample, the measurement on one-piece sample was also conducted for comparison of the sample without any crack. The results of the initial healing phase and discussions of its influencing factors are presented in Section 4. The results of the time dependent healing and discussions of its influencing factors are given in Section 5. 4. Initial healing results and discussions 4.1. Initial healing curve Fig. 3 shows the results of the initial healing tests. It can be observed that both the complex modulus of the one-piece sample and the two-piece sample increase when the gap thickness decreases. The one-piece sample shows a gradual increase of the complex modulus. And the two-piece sample exhibits a fast initial increase compared with the one-piece sample. It should be noted that during the analysis, the interaction between bitumen samples and the steel plates are assumed to be fully bonded. The development of the complex modulus is then interpreted as the response of the one-piece or the two-piece bitumen samples.

4.2.1. Compressive force and relaxation Fig. 4 shows the example of the monitored normal compressive force that developed during the gap thickness decrease. From the point of view of an energy balance, this compressive force in this test is a major energy inputs because the temperature kept constant. Within the setup, the compressive work either transfers into deformation of the samples or causes the closure of the crack. All of those can result in the increase of the complex modulus. Hence, it is believed that the compressive work is the main driving force for the increase of the complex modulus. It can also be noticed that even when the gap thickness keeps constant, relaxation still occurs inside the sample to release the residual normal compressive force, which might be an influencing factor of the initial healing curve. As a result, an initial healing test with a normal force which was fully relaxed (till compressive force of 0.1 N) was carried out and the results are shown in Fig. 5. It is interesting to note that the general trends of the initial healing samples with full relaxation are exactly the same as the samples under no relaxed condition, ignoring the height differences due to the variation of the sample thickness. 4.2.2. Gap thickness sensitivity For the second stage of the initial healing curve, a dramatic change of the complex modulus with only 0.5 mm gap thickness decrease is not expected. However, it explains why the time dependent healing test results are very sensitive to the sample thickness, which will be shown later in Section 5. 4.2.3. Geometry change In principle, the initial healing measurement results for the one-piece sample should be constant at each gap thickness. However, the decrease of the gap thickness resulted in a significant increase of the complex modulus. The reason could be: (a)

2713

J. Qiu et al. / Fuel 90 (2011) 2710–2720 Table 2 The TPH test procedures.

Phase 1 Initial healing test

Test on two-piece samples

Test on one-piece samples

Test samples Initial two-piece Initial two-piece trim

Test samples Initial one-piece Initial one-piece trim

Procedures A1. Stick two pieces of sample on the top and bottom plates, and keep test temperature at 20 °C

Procedures B1. Stick one-piece of sample on the top plate of the DSR, and keep the temperature constant at 20 °C B2. Decrease the gap to the condition that the sample fully touches the bottom plate

A2. Decrease the gap from 7000 lm to the condition that two pieces contact with each other with visual inspection using the camera inside of the temperature chamber A3. Start the complex modulus measurement applying a strain level of 0.00625% with frequency of 10 Hz for 3 min with a constant gap thickness. The strain level is chosen to be within the linear visco-elastic range and small enough to be not to interrupt the healing process A4. After the test, decrease the gap with each 200–500 lm and repeat the test procedure A3. Monitor the normal force development during the decrease A5. Trim the sample at the gap thickness of 5000 lm, 2000 lm, 1000 lm and 500 lm. Measure the complex modulus before and after trimming according to procedure A3 Phase 2 Time dependent healing test with gap constant control

Time dependent healing test with normal force constant control

B3–B6. The same as A3 to A5

Test samples TP-GC-a, TP-GC-b, TP-GC-c

Test samples OP-GC

Procedures C1. Same as A1 C2. Decrease the gap thickness of the parallel plates to a target gap thickness of 5.8 mm C3. Start the measurement applying a small strain level of 0.00625% at 10 Hz during 2 min followed by a rest period of 8 min C4. During the test, the gap thickness remains constant

Procedures: D1. Same as B1 D2. Decrease the gap to a gap thickness of 5.8 mm D3 and D4. Same as C3 and C4

Test samples TP-NF-a, TP-NF-b, TP-NF-c

Test samples OP-NF

Procedures E1–E3. The same as C1, C2 and C3 E4. During the test, the normal force remains constant at 0+/ 0.1 N

Procedures F1–F3. Same as D1–D3 F4. Same as E4

the orientation of bitumen molecules of the sample due to the shear force applied, which was reported in the low gap thickness range [31]; (b) the measurement itself, which generates more er-

ror at a higher gap scale (normally the recommended gap thickness is around 1–3 mm) [32]; (c) the change of the sample geometry.

Complex modulus [Pa]

1.00E+07

1.00E+06

Initial two-piece

1.00E+05

Initial one-piece

1.00E+04 4000

4500

5000

5500

6000

6500

Gap thickness [micrometer] Fig. 3. Initial healing curves with the decrease of the gap thickness.

7000

7500

J. Qiu et al. / Fuel 90 (2011) 2710–2720

Fig. 7 shows the influence of trimming on the two-piece sample. At a gap thickness between 6000 lm and 5000 lm, the change of the modulus on a two-piece sample due to trimming is less than the one-piece sample. However, it can also be observed that the increase of the trimming influences on the complex modulus of the two-piece sample at the low gap thickness range. At the second stage of the initial healing curve (from 6000 lm to 5000 lm), it seems that the geometry change during trimming has a great influence on the one-piece sample, but not too much influence on the two-piece sample. The reason could be the different geometry change mechanisms between the one-piece sample and the two-piece sample. For the one-piece sample, the complex modulus measures the response of the whole specimen. The mass surrounded the plate geometry would cause a big difference in the complex modulus measurement. For the two-piece sample, the complex modulus mostly measures the response of the weakest part, the shear modulus of the artificial crack between the two

Gap thickness [micrometer]

During the test, the gap decrease could cause the sample geometry to change, and part of the sample could exceed the plate geometry (25 mm in diameter). In order to investigate the influence of the sample geometry change on the initial healing measurement, a trimming action was performed. The difference in measured complex modulus before and after trimming was compared and the difference was referred to the geometry change effect. Fig. 6 shows the influence of trimming on the measured complex modulus of the one-piece sample. When the gap thickness decreases from 6000 lm to 5000 lm, the complex modulus of the untrimmed sample increases continuously. When trimming is applied to on the sample at a gap thickness of 5000 lm, the complex modulus of the trimmed sample becomes less than that of the untrimmed sample. The change in complex modulus due to trimming can also be observed at the gap thickness of 2000 lm, 1000 lm and 500 lm respectively, but with less difference.

6600

9

6500

8

6400

7

6300

6 Gap thickness

6200

5

normal force 6100

4

6000

3

5900

2

5800

1

5700

0

Normal force [N]

2714

-1

5600 4

6

8

12

10

14

Time [s] Fig. 4. Illustration of the compressive force during the decrease of the gap thickness.

Complex modulus [Pa]

1.00E+07

1.00E+06

1.00E+05 Initial two-piece Initial one-piece full relax Initial two-piece full relax

1.00E+04 4000

4500

5000

5500

6000

6500

Gap thickness [micrometer] Fig. 5. Influence of relaxation on the initial healing curves.

7000

7500

2715

J. Qiu et al. / Fuel 90 (2011) 2710–2720

7.00E+06

Complex modulus [Pa]

6.00E+06

5.00E+06

4.00E+06 one piece trimmed sample

3.00E+06

one piece untrimmed sample 6000-5000

2.00E+06 one piece untrimmed sample 5000-500

1.00E+06

0.00E+00 0

1000

2000

3000

4000

5000

6000

7000

Gap thickness [micrometer] Fig. 6. Influence of the geometry change on the one-piece initial healing sample.

6.00E+06

Complex modulus [Pa]

5.00E+06

4.00E+06

3.00E+06 two piece trimmed sample

2.00E+06

two piece untrimmed sample 5600-4500 two piece untrimmed sample 4500-1000

1.00E+06

0.00E+00 0

1000

2000

3000

4000

5000

6000

Gap thickness [micrometer] Fig. 7. Influence of the geometry change on the two-piece initial healing sample.

bitumen pieces. At the beginning of the gap decrease, the artificial crack closes. The deformation is less and the trimming influence on the complex modulus is also less. The response of the weakest part-the artificial crack rules the whole signal. After the two pieces fully touches each other, the same phenomenon of the one-piece sample can be observed. The healing percentage, which was defined by the ratio of the complex modulus of the two-piece bitumen specimen with that of the one-piece bitumen specimen over any given time point, cannot be used for quantifying the self healing effect because of the geometry change artefact on the one-piece sample. As a result, in order to eliminate the geometry change artefact on the complex modulus measurement of the one-piece specimen, the original value of the complex modulus development of the two-piece specimen will be considered for further time dependent healing comparison.

5. Time dependent healing results and discussions 5.1. Time dependent healing results 5.1.1. Time dependent healing with gap control (GC healing) Figs. 8 and 9 show examples of the time dependent healing test results with a constant gap thickness control (GC healing) for the two-piece and the one-piece samples. As shown in Fig. 8, the complex modulus of the two-piece sample increases slowly during time. A decrease of the normal force is also monitored throughout the test. This might be due to the relaxation of the residual normal force as shown in Fig. 4. A more detailed explanation of the decreasing normal force will be given in Section 5.2.1. Here, only the relative trends are compared. Fig. 9 shows the results of GC healing of a one-piece sample. The complex modulus keeps constant throughout the test. For the

2716

J. Qiu et al. / Fuel 90 (2011) 2710–2720

one-piece sample, a similar trend of the decreasing normal force is observed compared to the two-piece sample. Fig. 10 shows the results from the GC healing performed on three two-piece samples. The following can be observed:  Different tests gave different results. This is mainly due to the sensitivity of the complex modulus on the sample thickness.  In the first 20 min, a 5–10% increase of the complex modulus of the two-piece samples is observed. After that, a slight increase of the complex modulus is monitored over time.  When comparing the normal force with the level of the complex modulus, one can notice that a higher complex modulus is obtained at a lower normal force. 5.1.2. Time dependent healing with normal force control (NF healing) A time dependent healing test under the normal force controlled mode (NF healing) was also performed in this study. During the test, the normal force was kept constant automatically by adjusting the gap thickness between the upper and bottom plates. Figs. 11 and 12 show examples of the NF healing results of the two-piece sample and the one-piece sample, respectively.

For the two-piece sample, a rapid increase in complex modulus can be observed. The controlled normal force is around 0.12 N and 0.14 N, which is much smaller compared with the GC healing. Consequently, the gap thickness decreases from 5800 lm to 5500 lm to keep the normal force constant. A similar trend is observed for the one-piece sample (Fig. 12). Surprisingly, the gap thickness is decreased from 5800 lm to almost 4000 lm. The increase of the complex modulus over time can be observed, which is quite different when compared with the GC healing sample. The geometry change effect can also be observed due to the large change of the gap thickness. Similar to the initial healing results, only the original results of the complex modulus change of the two-piece samples are compared in order to eliminate the geometry change artefact on the one-piece sample. All the NF healing results are shown in Fig. 13, and the following can be observed:  The difference between the initial values of the time dependent curves can be observed. This is due to the thickness sensitivity effect. 0

2.00E+06 1.90E+06

Complex modulus [Pa]

-0.2 TP-GC-b Normal force

1.70E+06

-0.3

1.60E+06

-0.4

1.50E+06 1.40E+06

-0.5

1.30E+06 -0.6 1.20E+06

Decrease of normal force [N]

-0.1 TP-GC-b Complex modulus

1.80E+06

-0.7

1.10E+06

-0.8

1.00E+06 0

10

20

30

40

50

60

70

Healing time [min] Fig. 8. Test results of the two-piece GC healing sample TP-GC-b.

8.00E+06

0

OP-GC Complex modulus

-0.2

OP-GC Normal force

6.00E+06

-0.4

5.00E+06

-0.6

4.00E+06

-0.8

3.00E+06

-1

2.00E+06

-1.2

1.00E+06

-1.4 0

5

10

15

20

25

Healing time [min] Fig. 9. Test results of the one-piece GC healing sample OP-GC.

30

Decrease of normal force [N]

Complex modulus [Pa]

7.00E+06

2717

J. Qiu et al. / Fuel 90 (2011) 2710–2720

0 9.0E+06

Complex modulus [Pa]

-1 7.0E+06 -1.5

6.0E+06 5.0E+06

-2

4.0E+06

TP-GC-a TP-GC-b

3.0E+06

TP-GC-c

-2.5 -3

2.0E+06

Decrease of normal force [N]

-0.5

8.0E+06

-3.5

1.0E+06 1.0E+03 0

20

40

60

80

100

-4 140

120

Healing time [min] Fig. 10. Time dependent healing results with gap constant control.

6000

-0.05

-0.1

-0.15

-0.2

5000

3.0E+06

4000

3000

2.0E+06

2000

TP-NF-b Normal force

1.0E+06

TP-NF-b Complex modulus

-0.25

Complex modulus [Pa]

4.0E+06 Gap Thickness [micrometer]

Normal force [N]

5.0E+06

7000

0

1000

TP-NF-b Gap thickness

0.0E+00

0

-0.3 0

10

20

30

40

50

60

70

80

Healing time [min] Fig. 11. Test results of the two-piece NF healing sample TP-NF-b.

0.60

6000

1.20E+07

5000

1.00E+07

4000

8.00E+06

3000

6.00E+06

2000

4.00E+06

1000

2.00E+06

OP-NF Normal force

0.50

OP-NF Gap thickness

Gap Thickness [micrometer]

0.40

Normal force [N]

0.30 0.20 0.10 -0.10 -0.20 -0.30 -0.40

0.00E+00

0

0

10

20

30

40

50

60

70

Healing time [min] Fig. 12. Test results of the one-piece NF healing sample OP-NF.

80

Complex modulus [Pa]

OP-NF Complex modulus

2718

J. Qiu et al. / Fuel 90 (2011) 2710–2720

if the TPH test starts after full relaxation shown as the start point 1, the program will be zeroed with almost no normal force. If the test starts from the start point 2 (not enough time for relaxation), the normal force will be zeroed at zero line 2. As a result, a decreasing ‘‘negative’’ normal force would be expected during the test because of the remaining relaxation. During the self healing test described in this research, the test was started less than 1 min after pressing the two pieces of samples to a certain target gap thickness. Similar to start from the start point 2, the remaining relaxation can be expected. It explains the decrease of the decreasing ‘‘negative’’ normal force measured at the beginning of the GC healing test as shown in Figs. 8–10. However, in the NF healing test, the normal force control is active. After starting the measurement, the DSR machine indicates a decreasing ‘‘negative’’ normal force. Then the gap thickness is decreased in order to keep the constant normal force.

 The increase of the complex modulus over time is much faster compared to the GC healing (Fig. 10). For some samples, the development of the complex modulus with time is almost constant.  Different trends in gap decreases are shown in Fig. 12. The twopiece sample at a constant normal force of 0.12 N shows less gap decrease compared with the one-piece sample. 5.2. Normal force effects on time dependent healing 5.2.1. Discussion on normal force controlling mechanisms From the time healing test results mentioned above, a large influence of the normal force was observed. Hence, an equipment-based investigation was carried out to understand the normal force measuring and controlling mechanisms. According to the control manual of the DSR AR2000ex, the normal force control was realized by controlling the feedback signal from the normal force transducer located under the bottom plate as shown in Fig. 1. However, during the test, the normal force working function of the DSR is automatically zeroed when the test is started [32]. Fig. 14 shows the zero mechanism of the normal force of the DSR. After reaching a target gap thickness, the generated normal force starts to relax over time (also shown in Fig. 4). If the time is long enough, the normal force will relax to almost zero. Then

5.2.2. Normal force effect on NF healing results Through the NF healing test shows a much faster time dependent healing speed compared with the GC healing test, a decrease of the gap thickness is also observed in the NF healing test to keep the constant normal force. For the NF healing sample, the development of the complex modulus over time composes two parts: Time

1.0E+07

6000

5500

Complex modulus [Pa]

8.0E+06 7.0E+06

5000

6.0E+06 5.0E+06

4500

4.0E+06

TP-NF-a

4000

TP-NF-b

3.0E+06

TP-NF-c

2.0E+06

3500

1.0E+06 0.0E+00

3000 0

10

20

30

40

50

60

70

80

Healing time [min] Fig. 13. Time dependent healing results with normal force constant control.

Normal force

Start point2

Start point1 Zero line2 Zero line1 Time

Fig. 14. Illustration of working mechanism of the normal force.

Gap thickness [micrometer]

9.0E+06

2719

J. Qiu et al. / Fuel 90 (2011) 2710–2720

dependent healing with a constant gap thickness and additional healing due to gap decrease. This additional gap decrease can be related to the initial healing gap decrease. Fig. 15 shows the development of the complex modulus in relation to the gap thickness for both the NF healing and the initial healing results. The following comments can be made:  The three-stage curve can also be obtained in the NF healing test.  The development of the complex modulus of the NF healing two-piece samples follows the second stage of the initial healing curve.  The NF healing one-piece sample follows the third stage of the initial healing curve.

Fig. 16 shows the normalized three stages of the initial healing curves and the normal force healing curves. In order to avoid the thickness variation, the curve was shifted horizontally. The slope of the second stage of the NF healing in relation to the gap decrease is almost the same as the initial healing. Hence, it seems that most of the self healing signal through the NF healing is contributed by the gap decrease, which is caused by compressive force. As a result, the NF healing can be decomposed into the time dependent healing during a gap constant mode and the additional healing by compression. Since the contribution of the gap decrease part is dominant, one can suggest that compressive forces are much more beneficial for healing than healing times.

8.00E+06

Complex modulus [Pa]

7.00E+06 6.00E+06 5.00E+06 4.00E+06 3.00E+06 TP-NF-a

TP-NF-b

TP-NF-c

OP-NF

2.00E+06 1.00E+06

Initial two-piece

0.00E+00 0

1000

2000

3000

4000

5000

6000

7000

8000

7000

8000

Gap thickness [micrometer] Fig. 15. Development of the complex modulus vs. the gap thickness.

8.00E+06 7.00E+06

Complex modulus [Pa]

6.00E+06 5.00E+06

4.00E+06

TP-NF-a

3.00E+06

TP-NF-b TP-NF-c

2.00E+06

OP-NF 1.00E+06

Shiftted initial healing two-piece

0.00E+00 0

1000

2000

3000

4000

5000

6000

Gap thickness [micrometer] Fig. 16. Normalized three stages of the initial healing and the normal force healing.

2720

J. Qiu et al. / Fuel 90 (2011) 2710–2720

6. Conclusions and recommendations A detailed investigation of self healing behaviour of pure bitumen using a Dynamic Shear Rheometer has been reported in this paper. Based on the test data and analysis, the following conclusions can be made: 1. DSR is good for complex modulus measurements on bitumen. Since the healing may have an influence on the complex modulus obtained from the two-piece sample, DSR can be used to investigate the healing effect by using the development of the complex modulus as an indicator. 2. Initial healing phase: A unique three-stage curve of the initial healing modulus is obtained with the gap thickness decrease. 3. Time dependent healing phase: The normal force control time dependent healing shows much faster modulus improvement over time compared with the gap control time dependent healing. The normal force control healing can be decomposed into the time dependent healing with a gap constant control mode and additional healing by compression. It was also indicated that the compressive normal stress strongly promotes healing development. Furthermore, it was demonstrated that many factors can influence the complex modulus measurement results by using DSR parallel-plate geometry. Special attention must be paid to distinguish the real healing effect from those factors that may not really contribute to bitumen healing. Acknowledgements The authors appreciate the cooperation between Wuhan University of Technology and Delft University of Technology and express their wish to prolong and strengthen the current cooperation between the mentioned institutes of technology. The first author would like to thank the China Scholarship Council for the financial contributions. Special thanks to Dr. Gregory W Kamykowski from TA instrument for valuable discussions about the normal force control of the DSR equipment. Discussions and suggestions offered from Dr. Liantong Mo and Mr. Ad Pronk are highly appreciated. References [1] Bazin P, Saunier JB. Deformability, fatigue and healing properties of asphalt mixes. In: Proc 2nd int conference on the structural design of asphalt pavements, Ann Arbor, Michigan, 1967. p. 553–69. [2] Raithby KD, Sterling AB. The effect of rest periods on the fatigue performance of a hot-rolled asphalt under repeated loading. J Assoc Asphalt Paving Technol 1970;39:134–52. [3] Van Dijk W, Moreaud H, Quedeville A, Uge P. The fatigue of bitumen and bituminous mixes. In: Proc 3th int conference on the structural design of asphalt pavements, Ann Arbor, 1972. [4] Van der Zwaag S. Self healing materials: an alternative approach to 20 centuries of materials science. Dordrecht: Springer; 2007. [5] Francken L. Fatigue performance of a bituminous road mix under realistic best conditions. Transport Res Rec 1979;712:30–7.

[6] Little DN, Lytton RL, Williams AD, Chen CW. Microdamage healing in asphalt and asphalt concrete. Microdamage and microdamage healing, vol. I. Project summary report, FHWA-RD-98-144. Texas Transportation Institute, College Station, 2001. [7] Pronk AC. Partial healing model – curve fitting. Report WDWW-2000-047, DWW, Delft, 2000. [8] Hagos ET. The effect of aging on binder properties of porous asphalt concrete. PhD thesis of Delft University of Technology, Delft, The Netherlands, 2008. [9] Airey GD. Rheological properties of styrene butadiene styrene polymer modified road bitumens. Fuel 2003;82:1709–19. [10] Mo LT. Damage development in the adhesive zone and mortar of porous asphalt concrete. PhD thesis of Delft University of Technology, Delft, The Netherlands, 2010. [11] Mo LT, Huurman M, Woldekidan MF, Wu SP, Molenaar AAA. Investigation into material optimization and development for improved ravelling resistant porous asphalt concrete. Mater Des 2010;31:3194–206. [12] Huurman M, Mo LT. Lifetime optimisation tool, fatigue in mortar and adhesive zones; measurements, test interpretation and determination of model parameters. Research report 7-07-170-2. Laboratory of Road and Railway Engineering, Delft University of Technology, Delft, The Netherlands, 2007. [13] Kim YR, Little DN, Lytton RL. Use of dynamic mechanical analysis (DMA) to evaluate the fatigue and healing potential of asphalt binders in sand asphalt mixtures. J Assoc Asphalt Paving Technol 2002;71:176–99. [14] Kim YR, Little DN, Lytton RL. Fatigue and healing characterization of asphalt mixtures. J Mater Civil Eng 2003;15(1):75–83. [15] Lu X, Soenen H, Redelius P. Fatigue and healing characteristics of bitumens studied using dynamic shear rheometer. In: Proc of 6th RILEM symposium PTEBM’03, Zurich, 2003. p. 408–15. [16] Shen S, Chiu H, Huang H. Fatigue and healing in asphalt binders. In: Proc of Annual Meeting of Transportation Research Board, 2009. [17] Shen S, Chiu H, Huang H. Fatigue and healing in asphalt binders. J Mater Civil Eng 2010. dx.doi.org/10.1061/(ASCE)MT.1943-5533.0000080. [18] Van den bergh W, Molenaar AAA, van de Ven MFC, De Jonghe TC. The influence of aged binder on the healing factor of asphalt mixtures. In: Proc of AES– ATEMA’ 2009 third international conference on advances and trends in engineering materials and their applications, Montreal, Canada, 2009. [19] Phillips MC. Multi-step models for fatigue and healing, and binder properties involved in healing. In: Proc of Eurobitume workshop on performance related properties for bituminous binders, Luxembourg, 1998, no. 115. [20] Smith BJ. Hesp SAM. Crack pinning in asphalt mastic and concrete: effect of rest periods and polymer modifiers on the fatigue life. In: Proc 2nd Eurasphalt and Eurobitume Congress, Barcelona, 2000. p. 539–46. [21] Bahia H, Zhai H, Bonnetti K, Kose S. Non-linear viscoelastic and fatigue properties of asphalt binders. J Assoc Asphalt Paving Technol 1999;68:1–34. [22] Shan L, Tan Y, Underwood S, Kim YR. Application of thixotropy to analyze fatigue and healing characteristics of asphalt binder. In: Proc of Annual Meeting of Transportation Research Board, 2010. [23] Santagata E, Baglieri O, Dalmazzo D, Tsantilis L. Rheological and chemical investigation on the damage and healing properties of bituminous binders. J Assoc Asphalt Paving Technol 2009;78:567–96. [24] Bodin D, Soenen H, De la Roche C. Temperature effects in binder fatigue and healing tests. In: Proc of 3rd Eurasphalt and Eurobitume Congress, Vienna, 2004. [25] Bhasin A, Little DN, Bommavaram R, Vasconcelos K. A framework to quantify the effect of healing in bituminous materials using material properties. Road Mater Pave De, EATA 2008:219–42. [26] Bommavaram R, Bhasin A, Little DN. Determining intrinsic healing properties of asphalt binders: role of dynamic shear rheometer. Transport Res Rec 2009;2126:47–54. [27] Bommavaram R. Evaluation of healing in asphalt binders using dynamic shear rheometer and molecular modeling techniques. Msc thesis of Texas A&M University, USA, 2008. [28] Qiu J, Van de Ven MFC, Wu SP, Yu JY, Molenaar AAA. Investigating the self healing capability of bituminous binders. Road Mater Pave Des 2009;10:81–94 [SI]. [29] Qiu J. Self healing of asphalt mixes: literature review. Research report 7-08183-1, Delft University of Technology, 2008. [30] TA Instrument. AR-G2 2000ex Rheometers, Operator’s manual, 2006. [31] Zhai H, Bahia HU, Erickson S. Effect of film thickness on rheological behavior of asphalt binders. Transport Res Rec 2000;1728:7–14. [32] Personal communications with Dr. Kamykowski GW TA Instrument, October 27, 2009.