UNIAXIAL TENSILE TEST ON A NEW CEMENT COMPOSITE HAVING A HARDENING BEHAVIOUR

Transcription

1 UNIAXIAL TENSILE TEST ON A NEW CEMENT COMPOSITE HAVING A HARDENING BEHAVIOUR Claude BOULAY, Pierre ROSSI, Jean-Louis TAILHAN Laboratoire Central des Ponts et Chaussées (L.C.P.C.), France Abstract The LCPC patented a new cement composite (CEMTEC multiscale ), which have been developed to be stress hardening in tension and to have an ultra-high tensile strength. It is a multi-scale cement composite. One of the industrial applications aimed is related to thin slabs. Consequently, the research objective is to characterize the tensile behaviour of this material used in a thin bended slab. To achieve that, this tensile behaviour is measured from a bending test (easier to perform than an uniaxial tensile test) by an inverse approach. In order to validate the pertinence of this inverse approach, it is necessary to compare its results with those given by an uniaxial tensile test. So, an uniaxial test has been developed especially with this objective. This development has been done by using finite element analysis to determine the test specimen dimensions and the test set-up. The specimen retained is a dog bone specimen with 75 cm overall length. This paper presents in details the different steps which have permitted to conceive this uniaxial tensile test, and some experimental results obtained with it. These results confirms without ambiguity that this material is stress hardening in tension, very ductile, and has a uniaxial tensile strength of about 20 MPa. 1. Introduction Since a few years, new cement composites have been developed at the L.C.P.C. These materials are the direct implementation of the "Multi-Scale Concept" developed by Rossi [1]. The idea is to mix together short fibres with longer fibres in order to intervene, at the same time, on the material scale (increase of the tensile strength) and on the structure scale (bearing capacity and ductility).a Multi-Scale Cement Composite (MSCC) is then obtained. One can quote as a MSCC example, developed in the past [2] constituted by a mix of 5 % of straight cylindrical fibres (made of drawn steel) of 5 mm in length and 0,25 mm in diameter and 2 % of cylindrical hooked fibres (made of drawn 61

2 steel) of 25 mm in length and 0,3 mm in diameter. The uniaxial tensile behaviour of this MSCC is stress hardening and its mean tensile strength is about 15 MPa. The composite presented here, patented in the whole world by the LCPC in March 2001, is based on the same concept than the MSCC but with some evolutions : whereas the MSCC contains 7 % of 2 different metal fibre geometries, the CEMTEC contains 11 % of 3 of them. A vast study of this material started in 2000 over 4 years, including mechanical tests to characterize the various mechanical behaviours of the composite (static tests, fatigue tests, shock tests and so on), durability tests, tests on structural elements, and tests to optimize the manufacturing process (mixing and casting). This article describes the development of a specific uniaxial tensile test for this new composite and gives the first results obtained with this test. Out of the field of this article, it is planned to use the results of bending tests to rebuild the uniaxial tension behaviour by a reverse approach, the bending test being much easier to implement from an industrial point of view. Let's underline that the industrial application mainly aimed by this new composite is related to the two dimensional precast structures like slabs or shells. More precisely, this study has been performed in order to determine the experimental direct tensile behaviour which can be compared to those obtained by this reverse approach fed with the results of a bending test. The development of a new standard test was not the goal of this study. 2. Effect of the preferential orientation of fibres on the tensile behaviour of the FRCs It is significant to recall an aspect which conditions the distribution of fibres in a specimen. During the filling of a mould or a formwork, one generally observes zones in which the fibres have preferential orientations and zones where the orientations seem unspecified. These directions are obviously correlated with the FRC flows directions in the formwork during the casting and the vibration. One can suppose that, in a homogeneous concrete flow rate, viscosity being very high, a particle does not have tendency to pivot on itself but rather to be swept along at the same rate as concrete flow. In this case, the fibres do not present preferential orientation. One can also suppose that fibres rotations appear when rates gradients are established during the concrete flowing in the formwork. A long fibre will be more subjected to the rotations than a short fibre. Or rather, the fibres definitely longer than the size of the grains which surround it, will pivot easily because of a rate gradient of these particles. The fibres of the same size as the grains which surround it will not have a preferential orientation. The rate gradients appear close to the walls and the rebars. They are significant if the casting in a formwork is carried out in only one point. This explains why when one casts a beam (flat casting) like that of this study, the fibres close to the bottom of the mould present a mean orientation parallel to the bottom of the mould and according to the axis of the beam. For our specimens, this was carried out by gradually advancing from one end of the specimen to the other. In that case, the end of 62

3 the fibres close to the bottom of the mould is at almost a null speed while the other end is laid down by the concrete flow in a direction parallel to the bottom of the mould. The fibres located at the higher part of the beam do not easily rotate and their orientation is quasi unspecified. It should be noted that for the cement composite of this study, the shortest fibres are definitely longer than the coarser grains. So one can consider that all the fibres of this concrete are able to undergo a preferential orientation close to the bottom of the mould. Fig.1 : Fibres preferential orientation in the thickness of a beam and crack pattern related respectively to an uniaxial tensile test and a bending test. Figure 1 illustrates what precedes as for the fibres preferential orientation in a beam thickness. The more significant density of fibres in the bottom of the beam implies that the uniaxal tensile behaviour of its bottom part is higher than that of the total section. So, the CEMTEC uniaxial tensile behaviour determined from the uniaxial tensile test will be systematically under estimated compared to the uniaxial tensile behaviour determined from the bending test by an inverse approach. 3. Uniaxial tensile test development The objective of this test is to determine the uniaxial tensile behaviour of the CEMTEC when it is used in a thin slab. Deflection and compressive behaviours of the same material also used in thin slabs, with thickness not exceeding 50 mm, have already been the subject of an experimental study [3]. It was decided that the uniaxial tensile specimen thickness should not be higher than this value. As it is necessary that the fibres distribution is representative of that which exists within a slab, i.e. orthotropic, it was 63

4 decided that the width of the specimen, in the uniform tensile zone, was to be at least 4 times larger than the length of the largest fibre, in fact at least equal to 10 cm. Other criteria also intervened in the definition of the specimen geometry : the strain and stress fields in the zone of the specimen where the strain measurements are carried out must be most as uniform as possible, and over a sufficient length with respect to the length of the largest fibre: a measurement base of 20 cm was selected; the specimen should not crack in the zone located close to the interface between the specimen and the press, zone where the strain and the stress fields are not uniform. However, previous studies (see introduction) on cement composites with a matrix close to that of the studied material indicated that this matrix starts to crack for a tensile stress of about 8 MPa; in comparison, always with these previous studies, one can expect that the uniaxial tensile strength of the studied cement composite reaches 25 MPa, i.e. approximately 3 times the matrix tensile strength. According to the criteria and the above remarks, the specimen dimensions should respect the following conditions: it must have 5 cm thickness, 10 cm width, and 20 cm length, in the zone where the strain and the stress fields must be as uniform as possible. Moreover, it must have a 30 cm width at the level of the interface with the press (so that the stress is three times lower than that of the zone where strain measurements are carried out). Taking as a starting point the works of Do [4] and Behloul [5, 6], a dog bone specimen with 75 cm overall length was retained at this step of the study. The geometry of the specimen, glued to the press by using an adaptation block (made of aluminium), was optimized by using the finite elements code CESAR-LCPC. Taking into account the specimen shape and to avoid the generation of tensile stresses related to the material restrained shrinkage during its stay in the mould after the casting, particular cares, following those mentioned by Behloul [6], were taken (figure 2). The specimen geometry is also given in figure 2. Three moulds were manufactured to be able to run nine specimens in three batches. The mix design is given in table 1. Table 1. Mix design of the CEMTEC (kg/m 3 ) Cement Silica fume Sand Water Superplasticizer 44.0 Steel fibers Water/cement = Water/ binder = 0.16 Air entrained = 20 litres 64

5 Fig. 2-Specimen mould and geometry Fig. 3-Uniaxial test set-up Concerning the castings, the specimens were cast flat and were vibrated during the casting on a mobile plate. It is known that with the Ultra High Cement Composites, like the CEMTEC, the use of a thermal curing makes it possible to increase the mechanical performances of the matrix [7]. Also, in the present study, a heat treatment is applied which consists in placing the specimens in a drying oven at 90 C during 4 days, 48 hours after their release from the mould. After this cure the higher and the lower faces of each specimen were grounded with a surface grinder machine. Twenty minutes before the tests, the specimens were glued, with a methyl methacrylate resin, between two platens, made of aluminium alloy, rigidly fixed on the cross-heads of the testing machine. To ensure that no crack appears in the glued interfaces during the test, it was decided to add complementary supports placed on the specimen concave parts, these supports being connected to the specimen and to the aluminium platens via four 20 mm diameter prestressed rods (figure 3). Finite element calculations, still using CESAR-LCPC code, have been performed taking into account all the experimental set-up. They show that the tensile stress at the specimen/press interfaces reaches 8.5 MPa when it reaches 27 MPa in the specimen central part. So the interface is stiff, glued and pre stressed. 4. Experimental results Four specimens were tested following the experimental set-up described above. The jack displacement was controlled at a rate of 0.1 mm/min. The strain measurements are realized by using four transducers (LVDT) glued on each specimen face in its central part on a base of 200 mm. 65

6 Fig. 4 : Stress-strain curves given by the four transducers - Example related to one of the four specimens tested. Fig. 5 : Stress-strain curves related to the four specimens tested. In the figure 4 are presented the strain-stress curves given by the four transducers. It is a representative example of the four specimens tested. This figure 4 shows that the information given by the four transducers are similar and that we can consider the strain in the specimen central part as enough well homogenous. So, it is pertinent to determine the specimen tensile strain by considering the average value of the four displacements. Figure 5 gives the four strain-curves obtained with the four tested specimens. Only the stress pre-pick behaviour is considered. It confirms that material is a stress hardening in tension. Figure 6 presents an example of cracks pattern at the end of the test. It shows the existence of a multi-macrocracking. 66

7 Fig. 6-Craks pattern obtained at the end of the test. The characteristics of this new cement composite are gathered below : mean Young modulus : 55.5 GPa ; mean tensile strength : 20 MPa ; mean tensile maximal strain : ; 5. Conclusions In this article is presented the development of a new uniaxial tensile test to study a new stress hardening cement composite developed at the LCPC. The specimen geometry and the test-set up have been optimized to obtained homogeneous stress and strain fields in the specimen central part, where the strain measurement is performed. The direct tensile specimen dimensions were adapted in order to be representative of the composite behaviour in a thin slab. The result is a dog bone specimen of 75 cm in overall length. The central part is of 20 cm in length and 10 cm in width. The ends are of 30 cm in width. The specimen thickness is equal to 5 cm. Four specimens were tested. The principal results obtained are : The CEMTEC is a stress hardening material in tension; mean Young modulus : 55.5 GPa; mean tensile strength : 20 MPa ; mean tensile maximal strain : ; Due to preferential fibres orientation, these mechanical performances in uniaxial tension will systematically under estimate the tensile behaviour determined from the bending test by an inverse approach. 67

8 References 1. Rossi P., Acker P., and Malier Y., "Effect of steel fibres at two stages: the material and the structure," Materials and Structures, vol. 20, 1987, pp Rossi P., " High performance multimodal fibre reinforced cement composite (HPMFRCC) : the LCPC experience, " ACI Materials Journal, vol. 94, n 6, 1997, pp Rossi P., Arca A., Laurence O., Parant E., Fakhri P., "Comportement mécanique d'un nouveau composite cimentaire à écrouissage positif. I. Comportement en flexion," Bulletin des Laboratoires des Ponts et Chaussées, 2002, n 238, pp (in french). 4. Do M. T., "Fatigue des BHP, " PhD Thesis, Sherbrooke University, Canada, 1994 (in french). 5. Behloul M., "Définition d'une loi de comportement du BPR, " Annales de l'itbtp, n 532, 1995, pp (in french). 6. Behloul M., "Analyse et modélisation du comportement d'un matériau à matrice cimentaire fibrée à ultra hautes performances - (béton de poudres réactives) - Du matériau à la structure," PhD Thesis, Ecole Normale Supérieure de Cachan, France, 1996 (in french). 7. Richard P., Cheyrezy M., Les bétons de poudres réactives, Annales de l I.T.B.T.P., 532, 1995, pp (in french). 68