An extended model to predict the compressive, tensile and flexural strengths of HPFRCs and UHPFRCs: definition and experimental validation

High manufacturing costs of UHPFRC and expensive and time-consuming tests performed to understand the mechanical response under loading restrict still its wider applications in the field of the structural engineering. Predictive models can be useful to reduce the number of requested tests and to optimize the amount of compounds of the mixture, for example detecting the minimal dosage of fibers necessary to attain a given tensile strength and toughness as well. Currently, not many predictive models do exist and one of the most recent, developed in order to estimate the compressive and tensile responses of HPFRCs, was not notably suitable for UHPFRCs. The main purpose of this work concerns the extension of such a model, in order to predict the mechanical response (in flexion as well) of a given HPFRC/UHPFRC for any change of matrix and fiber properties. Theoretical results were compared with experimental data, thus confirming some shortcomings of the previous model. Once the matrix and fiber properties of a marked UHPFRC were selected, the extended model was used to predict the tensile and flexural bending responses of a full scale UHPFRC structural beam, showing good agreement with the experimental results.

mechanics; Mechanical testing the mechanical performances of a full scale structural beam. 48 The present work is organized as follows: a description about the materials 49 and test methods adopted in the experimental program are provided in Sec-50 tion 2; in Section 3 the most relevant results of the experimental investigation 51 are highlighted; the extended model is presented in Section 4; conclusions are 52 drawn in Section 5. 53 54 The empirical previous model developed in [35] was here extended in 55 order to predict the compressive, tensile and flexural bending strengths of 56 a given HPFRC/UHPFRC for any change of matrix and fiber properties. 57 Experimental data of strength tests recorded on both marked UHPFRC (la- 58 beled hereafter UHPFRC-A) and marked HPC series -investigated in [35]- 59 by varying fiber properties, were taken into account for this purpose. In   The geometry of the full scale beam and the position of the drilled specimens 84 are illustrated in Fig. 3. As control, a UHPFRC-B series of 100 × 100 × 100 85 mm cube and dog-bone shaped specimens were made in lab according to 86 standard conditions. A series of specimens without fibers was also made. 87 In order to understand the scatter in strength between cores and cast (cylin-88 der) specimens a proper study is discussed afterwards, by using a marked 89 UHPFRC (labeled hereafter UHPFRC-C), see Table 1.

90
The mix design of UHPFRCs here investigated was provided by the man-91 ufacturer ( Table 2), even though for UHPFRC-A series different kinds of 92 fibers and dosages were investigated. Typically, marked UHPFRCs include 93 an optimized gradation of granular matter to obtain a high packing density.

94
A high intensity mixer was used to ensure the mix homogeneity. The fibers 95 were inserted in the mix in order to obtain a good dispersion and minimizing 96 the risk of fiber balling. The specimens were made and cured in compliance 97 to the European standards [58].

99
The amount of fibers within the UHPFRC-A (cubic and dog-bone-shaped) 100 specimens was investigated by measuring the hardened-state density in all 101 series according to [60]. Standard uniaxial tensile tests on dog-bone-shaped     is to use multiple fibers bonded together with water-soluble adhesive [5]. 126 This solution was adopted by replacing HF85 with such a kind of fibers (HF

Influence of fibers on compressive strength (UHPFRC-A series)
It is observed that fibers increased the compressive strength f c until 39%  Table 3.

222
In Fig.9a the deformation of the sensors is expressed in micro strain µ , while 223 in Fig.9b the temperatures are shown for the central and the side sensors.

224
The temperature sensors seem to demonstrate that the water used for cooling   for a faster hydration) while at 28 days they test lower of more than 10%, 241 thus confirming that the strength retrogression occurred also in this case 242 (Fig.10a).

243
In order to confirm that size of specimens and molds type were not an issue,      (Table 3 and Table 3 in [35]) and predicted values of compressive, uniaxial and residual tensile strengths by the extended model (a-f) and the model in [35] (g-l) Figure 13: Correlation between experimental data of a, b, c ( [36]), d (UHPFRC-B series) and theoretical results both of the extended model and previous model developed in [35] parabola-rectangular stress-strain law in compression and bi-linear stress-284 strain laws in tension defined according to [52] and [57] were adopted in the

289
The comparison with experimental data was observed in Fig.13. It is worth