Influence of start temperature on tensile stress testing of restrained asphalt concrete specimens

This paper presents the result of the Tensile Stress Restrained Specimen Tests on specimens of asphalt concrete AC 8 with bitumen 50/70 depending on two starting temperatures of +10 °C and +20 °C. The analysis of the results shows that there is only a non significant difference between these two start temperatures and the results are within the limit of standard precision. On the other hand, there is a difference between the results of two laboratory TU Wien and ZAG Ljubljana, who carried out the tests in accordance with EN 12697-46 at start temperature +20 °C


Introduction
In nature, most substances including asphalt expand when they are heated and contract when they are cooled.If the contraction due to cooling is prevented with falling temperatures increasing tensile stresses in the asphalt material will be generated, which can lead to fracture (micro-cracking in the binder matrix) if the maximum tensile strength is reached [1].Tensile Stress Restrained Specimen Tests (TSRST) simulate the condition of asphalt pavement at low temperatures, where the resulting thermally induced tensile stresses, called cryogenic stress, primarily reflect as transverse cracks spaced at 3 to 5 m [2].The [3] concept of tensile stress, tensile strength and tensile strength reserve is shown in Fig. 1.The thermally inducted (cryogenic) stress in asphalt specimen gradually increases as temperature decreases, until the specimen fractures.At the break point, the stress reaches its maximum value -the fracture stress σ cry,fracture at fracture temperature T failure (hereinafter T f ).At lower temperatures the slope of the stress-temperature curve dS/dT becomes constant, and the curve is linear (elastic behaviour).The transition temperature T u divides the curve into two parts -relaxation zone and elastic zone (non-relaxation) and tangent point of intersection T TS is intersection between tangent of the stress-temperature curve at elastic zone and relaxation zone.The bitumen in asphalt specimen becomes stiffer when the temperature approaches the transition temperature and the thermally inducted stresses are not relaxed below this temperature [4].The standard EN 12697-46 recommends to start the test at a temperature of T 0 = +20 °C.At Vienna University of Technology (TU Wien) the TSRST are traditionally carried out at start temperature of T 0 = +10 °C.In the study presented in this paper, it was investigated whether the start temperature T 0 at the TSRST has an influence on the results at low temperatures.Therefore, TSRST was carried out according to standard EN 12697-46 on asphalt concrete AC 8 at starting temperature T 0 = +10 °C and = +20 °C.TSRST was carried out at ZAG Ljubljana Institutes on the asphalt samples with 6.2 m.-% content of bitumen to compare with results of TU Wien.The both Institutes used different equipment manufacturer for TSRST.ZAG Ljubljana Institutes carried out the TSRST on compression-puller manufacturer named "Frank" and TU Wien used the equipment of manufacturer "Wille Geotechnik" from Germany.

Test procedure and Equipment
For the TSRST, the asphalt concrete specimen is mounted in a load frame, which is enclosed in a cooling chamber.During the experiment the length of the specimen is kept constant and the temperature is decreased with a constant cooling rate of dT = -10 °C/h.Any movement of the specimen as consequence of thermal shrinkage is monitored by LVDTs, activating a screw jack, to keep the specimen at its original length.This process continues until the tensile stress exceeds the tensile strength and, hence, the specimen fails due to cracking.In standard EN 12697-46, it is recommended to start the test at a temperature of T 0 = +20 °C.The testing machine consists of a load frame, a screw jack, a climate chamber with temperature controller, 4 LVDTs, a specimenalignment stand and a computer data acquisition and control system [5,6].The climate chamber allows to control the temperatures within T = ±40 °C (TU Wien accuracy of ±0.1 °C and ZAG accuracy of ±0.5 °C).The LVDTs are placed outside of the climate chamber (accuracy of 0.2 %).The top and bottom plates and the measurement rods are made of invar steel to avoid a strong influence of temperature changes on the measurement device [5]. Figure 2 shows TSRST equipment of the TU Wien and illustration of setup of the employed testing equipment.Influence of start temperature on tensile stress testing of restrained asphalt concrete specimens results derived from TSRST on three samples (duplicates) by the same operator shall be considered suspected if they differ by more than 2 °C of the failure temperature and 0.5MPa of the failure stress.The precision highly depends on the range of void content of the samples of the asphalt mixture.An illustration of the test procedure of TSRST is given in Figure 3.

Material
The tests were carried out on asphalt concrete 0/8 mm (AC 8).The asphalt mixture and test specimens were prepared in the laboratory ZAG Ljubljana.The asphalt mixtures have 6.2 m.-% of bitumen content.Stone fractions used for the stone aggregate mixture are as follows: filler aggregate (grain under 0.125mm) from Stahovica (limestone), mineral aggregate 0/2, 2/4 and 4/8 mm from Ljubešćica (silicate) and for binder we used paving grade bitumen 50/70 by MOL (Hungary).Figure 4 presents grain size distribution for this asphalt mixture 0/8 mm (AC 8).The bitumen and aggregates were heated to prior to mixing at temperature T = 150 °C, for around 0.5 h.Also, prior to compacting, the mixture was conditioned at T = 150 °C, for 0.5 h.For TSRST tests we used a rectangular specimen with cross section dimensions 40 x 40 mm 2 and a length of 160 mm.After trimming, the specimens were conditioned at room temperature T = 20±2 °C and tested in a few days.Table 1 shows properties of used fresh and extracted with trichloroethylene Infratest Asphalt Analyzer (EN 12697-1) paving grade bitumen 50/70.Table 2 presents the results of some basic tests of asphalt mixtures.

Comparison between the two starting temperatures
The results obtained from testing of asphalt concrete AC 8, used for wearing courses on pavements, are presented in Table 3.Three specimens K348A-C have been tested at start test temperature T 0 = +10 °C and the remaining three asphalt samples K348D-G at T 0 = +20 °C.Statistically speaking, all the results of the failure stress showed that the spread between the results slightly above the permitted limit of 0.5 MPa by the standard EN 12697-46.At failure temperature, the results are within the requirements by the standard (<2 °C).If results are looked separately by the start temperature T 0 = +10 °C and T 0 = +20 °C, it can be seen that both cases do not exceed the requirements.Influence of start temperature on tensile stress testing of restrained asphalt concrete specimens zone follow the same curve.In the elastic zone curves are slightly apart, but within expectations.The results of failure stress and failure temperature are graphically presented in Figure 7b, where it is seen that the results are comparable.An asphalt mixture, up to a certain temperature is capable of internal movements in the structure without causing a visible change on the outside forms (relaxation phenomenon).Between temperature of +10 °C and +20 °C, these relaxation characteristics of the mixture reduce inducted thermal stress in asphalt specimens and thus that differences between TSRST at start temperatures of +10 °C and +20 °C are negligible.
If a body has one dimension which is much larger than the other two, the two smaller dimensions can be ignored and only the expansion of the length can be looked at.Change of the body length Δx with temperatures change is defined as: where is x initial body length, ΔT the temperature change and α T thermal expansion coefficient of the length.If input in the equation ( 1) is given as x =161mm (length of the asphalt specimen), ΔT = +20 °C -(+10 °C) = 10°C and α T = 2.2 x 10 -5 [5] as the thermal expansion coefficient of asphalt concrete then change of the specimen length is Δx = 0.0354 mm.According to the required resolution of the LVDTs in EN 12697-46 (0.5 μm = 0.0005 mm), this displacement should be measured and controlled the specimen original length by the test machine.Therefore, equipment does not effect on the results at start temperature T 0 = +10 °C.

Comparison between the two laboratories
Figure 8 shows the stress-temperature curves of TSRST, which were carried out on prismatic asphalt concrete samples at ZAG Ljubljana (red curve) and TU Wien (blue curve).The curves split in two parts between the temperatures around 5 °C.At colder temperature both curves runs parallel.At tensile stress of 1.0 MPa the temperature difference between curves is between 5-6 °C.The mean curve TU Wien is average of three curves (K384D, K384F and K384G) and the equation reads as follows Mean curve ZAG is average of curve 001, 002 and 003 with equation: y 2 (x) = -2E-09x 6 -8E-09x 5 + 1E-06x 4 -8E05x 3 + 0,0029x 2 -0,0365x + 0,1962 (3) If we want to get the tangent line equation y(x) = kx + n to a curve at a given point in the elastic zone, it is necessary to derivative the curve function y(x).In this study the tangent is calculated at the brake point V1 (T f = -31.09°C, σ cry,f = 5.19 MPa) for TU Wien and V2 (T f = -26 °C, σ cry,f = 4.26 MPa) for ZAG.The equation of both tangents is shown on Figure 8.The intersection of the tangent and the curve is the transition temperature T u .The transition temperature is T u1 = -27.2°C for TU Wien and T u2 = -24.1 °C for ZAG.Therefore, the relaxation zone at TU Wien is longer as ZAG.
In the elastic zone the curve slope of the results derived at TU Wien (dS/dT = 0.4035) is little higher that at ZAG (dS/ dT = 0.2955).Figure 9 shows the results of TSRST test at start temperature T 0 = +20 °C for TU WIEN and ZAG.The difference of the average temperature at failure T f between TU Wien and ZAG is 5 °C and is outside the permitted range in accordance with EN 12697-46 (<2 °C).The TU Wien samples have higher tensile stress as ZAG (for 0.57 MPa), which is outside the permitted range (<0.5 MPa).The results of the TSRST at TU Wien have better resistance to cracking at low temperatures.All of these differences could be due to various factors as preparing and gluing samples, precision or resolution of LVDTs or differences in the test machine.

Conclusion
The result of Tensile Stress Restrained Specimen Tests (TSRST) on specimens of asphalt concrete AC 8 surf with bitumen 50/70 depending of two starting temperatures of T 0 = +10 °C and +20 °C is shown in this paper.The analysis of the results TSRST at TU Wien shows that there is no influence between these two starting temperatures and the results are within the limit of precision given by the standard EN 12697-46.At the University of Nevada Reno found that effect of different starting temperatures (T 0 = +5 °C and +20 °C) is smaller than effect of different cooling rates [7].In the relaxation zone both curves (T 0 = +10 °C and +20 °C) in the stress-temperature diagram are congruent.In the elastic zone curves are slightly apart, but within expectations.Since at temperature of +10 °C and +20 °C the viscous component in asphalt mixture is dominates and the relaxation characteristics of the asphalt mixture reduce inducted thermal stress, the difference between TSRST at starting temperatures of +10 °C and +20 °C is negligible.The results of TSRST, comparison between ZAG Ljubljana and TU Wien, shows that there is a considerable difference, although both Institutes carried out TSRST in accordance to the standard EN 12697-46.In this case the both curves TU Wien and ZAG run apart at temperature around 5 °C, but at colder temperature run parallel.At a tensile stress of 1.0 MPa the temperature difference between curves is 5-6 °C.
The relaxation zone at TU Wien is longer as ZAG and in the elastic zone the slope of curve is little higher at TU Wien.All of these differences could be due to various factors (etc.gluing samples, equipment).

Figure 2 .
Figure 2. a) TSRST equipment in TU Wien; b) illustration of setup of the employed testing equipment The results of a TSRST are the progression of the temperaturedependent cryogenic stress σ cry (T) [MPa], the failure stress σ cry,failure [MPa] and the failure temperature T failure [°C].The

Figure 4 .
Figure 4. Grain size distribution of AC 8

Figure 5 .
a and 6.a show stress-temperature diagram of the TSRST at start test temperature T 0 = +10 and +20 °C carried out in TU Wien.The both curves have smooth line.Figure 5b and 6b presents results of failure stress and failure temperature of the TSRST test at start test temperature T 0 = +10 and +20 °C.Both stress -temperature curves at start test temperature T 0 = +10 °C (blue line) and +20 °C (red line) are presented in Figure 7a.Practically, the both curves in the relaxation

Figure 5 .Figure 6 .
Figure 5. Results of TSRST test at start temperature T 0 = +10 °C: a) cryogenic stress σ cry (T); b) failure stress σ cry , failure and the failure temperature T failure