False Failure in Flexural Fatigue Tests

Flexural fatigue tests are typically run under displacement or in a straincontrolled mode. In these tests, either the oscillatory displacement amplitude or strain amplitude applied to the bottom of the specimen is kept constant. The evolution of loading required to cause fatigue is then measured. Load amplitude decreases with the number of cycles, and the specimen is considered to have failed when the load is half its initial value. This failure criterion may be erroneous when non-fragile fracture mixtures prepared with high bitumen contents or modified binders are tested. In these cases, mixtures exhibit a visco-plastic behaviour and increasingly less stress is necessary to cause strain without cracking. Mixtures are hardly deteriorated when the fatigue failure is determined, and may be subjected to a larger number of load repetitions. It is then recommended to control the evolution of loading more effectively and regard as valid only tests where load decreases sharply to very low levels, making sure that the three stages in the fatigue process have occurred. Other researchers, Breysse et al. (2003, 2004) have also studied and modelled the influence of rest time on damage during fatigue tests and have shown the potential precariousness of the healing. The fatigue failure process of bituminous mixes has been studied in the Road Research Laboratory of the Technical University of Catalonia through the classical theory of fatigue failure and special attention has been paid to the strain evolution in the fatigue fracture region. Flexural and direct tensile fatigue tests were conducted to determine the strain evolution during the fatigue process of a series of bituminous mixtures with the aim of determining whether there is or not a certain level of permanent strain at which the mix fails due to fatigue process, irrespective of the stress or strain level applied. 2 MATERIALS AND METHODS Different semidense bituminous mixtures with a maximum aggregate size of 20 mm have been tested through three and four point bending beam tests and direct tensile fatigue test. Although they have been prepared with different RAP contents and different bitumen types, the grading was kept constant (Table 1), as well as total bitumen content (4.5% by mass of aggregate). Mixtures with 30%RAP used 80/100 penetration bitumen, mixtures with 50%RAP used 150/200 penetration bitumen and mixture without RAP were fabricated with 60/70 penetration bitumen. The average grading of the aggregates is summarised in Table 1. Table 1. Average grading for the bituminous mixtures studied Sieve Size (mm) Passing (%) 25 100 20 91.8 12.5 71.3 8 59.6 4 41.8 2 28.8 0.5 14.4 0.25 10.3 0.125 7.1 0.063 5.3 Two flexural fatigue tests have been used: four and three-point bending beam tests (the last one standardized in Spain), and also a dynamic direct tensile test developed in the Road Research Laboratory of the Technical University of Catalonia, with the objective of comparing the results obtained with the bending beam tests and another fatigue test type. The tests have been carried out at different temperatures: 5, 20 and 35oC, in order to analyse different mixture performances: stiff and fragile behaviour at low temperatures, flexible and ductile behaviour at high temperatures. Four point bending beam was carried out according to UNE-EN 12697-24, Annex D. Figure 1 shows the four point bending equipment, composed of two inner and two outer clamps symmetrically placed. The two outer clamps keep the beam fixed and the two inner clamps are loaded to create a constant moment. Figure 1. Four point bending beam test. Three point bending beam was carried out according to Spanish Standard NLT-350 (or UNEEN 12697-24, Annex C), with a prismatic specimen laid on its ends and fixed in its centre. Figure 2 shows the anchoring devices for the specimen testing. Figure 2. Anchoring devices for the specimen testing. The dynamic bending beam tests (both three and four point beam tests) consist in subjecting a prismatic specimen to a time-variable displacement according to the following law: ) 2 sin( . 0 ft D D π = (1) where D = displacement at moment t; 2D0 = total amplitude of the displacement function; f = wave frequency; and t = time. Direct tensile test was developed at the Road Research Laboratory of the Technical University of Catalonia and consists in subjecting a prismatic specimen to tensile stress. Prismatic specimens can be obtained from laboratory-made prismatic or cylindrical specimens of different sizes depending on the compaction equipment used. They must be properly sawn to achieve approximate dimensions of 150x50x50 mm. Alternatively, they can be obtained by sawing cylindrical cores extracted directly from the pavement layer. In both cases, a small 5 mm indentation is made at both sides of the central section of the specimen. A metallic support is stuck on the bases of the specimen so that clamps placed on each of the press pistons are fixed to the specimen. In this way, tensile stress can be applied to the specimen, Figure 3a. During the specimen testing, the variation of strain produced in the mixture is recorded with one or two extensometers placed on one or both indented sides of the specimen, respectively, Figure 3b. (a) (b) Figure 3. Direct Tensile Test, (a) photo, (b) scheme. Under stress controlled mode, tests were carried out at a frequency of 10 Hz and a cyclic loading is applied according to a sinusoidal function expressed in equation (2). During the test, strain evolution with the load applications is registered. A ft F F + = ) 2 sin( . 0 π (2) min max 0 2 F F F − = (3) 2 / ) ( min max min F F F A − + = (4) where F = load at moment t; 2F0 = total amplitude of the load function; Fmax= maximum load; Fmin= minimum load; f = wave frequency; and t = time. 3. RESULTS AND DISCUSSION When the beam is subjected to a four point bending test, its modulus decreases with the increase of the number of cycles applied. The value of the initial modulus is calculated from the measured values of force, displacement and phase lag after the hundredth cycle. According to the classical fatigue failure criterion, the fatigue test continues until the modulus drops to half its initial value or until the specimen breaks. However, if the test is stopped at that moment of 50% modulus reduction (and the specimen is not broken), and the strain amplitude is increased, the specimen will behave as if it is not broken, since the modulus turns to be high and the test will continue until the modulus reaches half its initial value. Figure 4 shows the load evolution, Press piston Extensometers Fixed base Bituminous mixture specimen Piston movement, static mode directly related to modulus evolution, where the specimen was tested according to the above mentioned procedure. Figure 4. Load evolution with the number of cycles. Four-point Bending Beam Test at 20oC, semidense bituminous mixture. So, the authors put forward that a mistake can be made if the specimen failure is considered when its modulus is reduced to a 50% of its initial value and the strain evolution is not taken into account, since the specimen may be considered to have failed much before it actually has. Flexural bending beam tests present this problem when they are performed at displacementcontrolled mode, especially when deformable mixtures containing polymer-modified bitumens or high bitumen contents are tested. In order to show this fact, and considering that the four point bending beam test does not allow registering the strain evolution at the bottom of the specimen, the authors use the three point bending beam test, where an extensometer can be fixed to the face of the beam, Figure 5. Figure 5. Three-point Bending Beam Test. Extensometer placed to measure the strain evolution. The results obtained from the test conducted in the dynamic mode reveal that, for each mixture type, there is a strain from which the fatigue process proceeds very rapidly. This strain, which we have called “critical strain”, is independent of the stress state to which specimens are subjected during the fatigue process. That is, if a high stress is applied, the initial strain will be greater and will increase with each load application until the critical strain level is reached. At this point, the fatigue process will speed up and the strain level produced in each cycle will increase until the material cracks. The initial strain will be lower if a smaller load is applied, 0 50 100 150 200 250