Pull-out tests have been used by various researchers in order to understand the bond between fibres and concrete. Most of the studies available in the literature concern the short term behaviour of steel fibres [13-19] but it is possible to find also studies on polymeric fibres [20-24].
To the authors knowledge the only experimental results published based on long-term pull out tests are those by Babafemi and Boshoff  and Nieuwoudt et al. , who used test setup in which the sustained load was applied to single fibres bonded in concrete using free hanging weights. The pull-out displacement was measured optically by taking digital images with a microscope.
Pull-out tests are attractive because they might allow a better understanding of the bond behaviour, even though using information on the long-term behaviour obtained from these tests in order to predict the behaviour of cross sections might be complicated by random fibre orientation and anchorage lengths.
Uniaxial Tension Tests
Van Mier and van Vliet  provide an overview on uniaxial tension tests procedures on quasi-brittle materials in general. The heterogeneity of these materials leads to various issues such as the need of large specimens, and the development of secondary flexural moments, causing redistribution of stresses. The experience of uniaxial tension tests on FRCs is in general limited. Plizzari et al. [27, 28] used uniaxial tension tests in order to investigate the behaviour of Steel Fibre Reinforced Concretes (SFRCs) under cyclic loads. Li et al.  investigated the uniaxial behaviour of SFRC and Macro Synthetic Fibre Reinforced Concrete (MSFRC) specimens with fibre volume fractions from 2 to 6 %. RILEM TC-162  issued guidelines for uniaxial tension tests based on notched cylindrical specimens and fixed end plates. These guidelines were then used in a round robin test program [31, 32] after which it was concluded that significant differences resulted from different testing labs. The authors suggested that they could be due to different stiffness in the setups. A large variability of the results was observed and was suggested that it was due to material variation, even if it was not possible to analyse whether the small cross section size of the cylinders played an important role. In spite of some difficulties encountered during test executions the authors concluded that the uniaxial test was robust. Using the same experimental setup Barragan et al.  performed a parametric study on specimens containing steel fibres and analysed the effects of various parameters such as the depth of the notch and the slenderness of the cylinders. They also investigated the effect of fibre orientation by considering cast and cored specimens, these later being cored from prims in different directions with respect to the prims axis. They concluded that specimens are to be cored in order to obtain representative fibre orientations, especially if they are to be compared with bending tests. This was confirmed also by Zhao et al.  and Buratti and Mazzotti .
Zerguini and Rossi  carried out uniaxial tension tests on SFRCS considering notched cylindrical specimens with different diameters (68, 100, and 150 mm) in order to investigate the sensitivity of test results to this parameter. Fibre dosages from 54 to 100 kg/m3 where used. The authors concluded that no significant dependence of either the average post-cracking energy in uniaxial tension or the dispersion relative to this characteristic with respect to specimen dimension could be identified, although it should be noticed that high fibre dosages were used. Sorelli et al.  carried out tests with freely rotating platens on prismatic specimens.
The literature on long-terms uniaxial tension tests is even more limited. Zhao et al.  performed uniaxial creep tests on notched cylinders. The setup used was composed of stiff steel plates epoxied at the ends of the cylinders which were connected by three loading bars used both to apply and to measure (using strain gauges) the tension force. The load was then applied by turning the fastening bolts on the loading bars. During the pre-cracking phase, the three loading bolds were fastened in such a way that the crack opening, which was measured using LVDTs in three positions at 120° around the circumference, was as uniform as possible. After pre-cracking the specimens were then tested under long-term loads using a lever based system. In this phase of the tests the rotation of the specimen ends was not blocked. Babafemi and Boshoff  carried out uniaxial tension tests on notched MSFRC prismatic specimens. Pre cracking was carried out in displacement control and then the specimens were loaded using a lever system. Buratti and Mazzotti  proposed a test procedure on notched cylinders on which the specimen ends are allowed to rotate, both in the pre-cracking and in the long-term test phase, by means of a spherical joint. During the pre-cracking phase, the authors identified specimens with anomalous rotations (e.g. parts in compression) which were then not used for long-term tests . Long-term tests were then carried out on a chain of three specimens loaded using a lever frame (Fig. 1).
In all the cases cited above the typical testing procedure can be summarized as follows (see Fig. 2). The specimen if first pre-cracked up to a defined crack-opening (O-A) at which the residual strength fR is measured. The average crack opening is normally considered to quantify the crack opening. Then the specimen it is unloaded (A-B) and transferred to the long-term testing frame. Since pre-cracking is normally carried out at very low crack opening rates, and therefore creep deformations develop during this stage, delayed deformations are in part recovered in the
Fig. 1 Experimental setup used by the authors for long-term uniaxial tension tests 
unloaded state (B-C). There is yet no agreement in the scientific community on the significance of these deformations and on whether they should be monitored. The specimen is then reloaded up to a fraction of the residual stress measured during pre-cracking (C-D), = afR, where a indicates the creep load ratio. The time tL in
Fig. 2 indicates the reloading time which should be as short as possible, in order to limit the interference between instantaneous and creep deformations. During the long-term tests deformations will increase at a constant load (D-E). At the end of the long term test the specimen is unloaded (E-F) and part of the long-term deformation is recovered (F-G). Finally, the specimen might be reloaded to failure in a short term test (G-H). It is worth noticing that tertiary creep (leading to failure) might be observed during the long-term tests. These tests are normally carried out in controlled environmental conditions.
In the authors’ opinion, even though testing protocols exist for short term tests , long term tests present some specific problems. In fact, during long-term tests it is in general not possible to block the rotation of the specimen ends. If specimens are pre-cracked using fixed ends an inconsistent behaviour might be observed during long-terms tests, because of different secondary moments. Furthermore, very large rotations can occur during the long-term tests, leading in some case to compression forces in a portion part of the cracked section. This is more likely to occur when the number of fibres crossing the crack surfaces is limited [34, 37]. Buratti and Mazzotti also observed that uniaxial tension tests, because of the small cross-section size of the specimens, tend to exhibit, on FRCs with low fibre dosages, a scatter of the results which is much higher that bending tests.
Uniaxial tension test, even if more complicated to carry out than bending tests (see Sect. 4), are, to the Authors’ opinion an interesting option because, if properly executed, they allow to study the average long term behaviour in tension and therefore their interpretation is more straightforward with respect to bending tests (see Sect. 4).
Fig. 2 Typical behaviour observed, in a softening FRC, in long-term tests in cracked conditions, in terms of nominal-stress versus crack opening (top panel) and crack opening versus time. COD indicates the general crack opening displacement and might correspond to either CMOD, CTOD or other parameters depending on the test setup