Keywords

1 Introduction

In real life structures, shrinkage of the concrete paste is restrained internally by the aggregates and reinforcement and externally by the ground or adjacent structural elements. The degree of restraint cannot be quantified accurately as it depends on the type of the application, the member location and the environmental conditions (Banthia et al. 1996; Gilbert 2016). However, there have been several attempts in the literature to simulate the restrained shrinkage of the concrete. An active system (Bloom and Bentur 1995; Paillere et al. 1989; Altoubat and Lange 2001) was used to restrain concrete shrinkage by fixing one end of the prism and keeping the second end free to move while it was attached to a special gauge. The gauge was used to maintain the applied force to the free end to return it to its original place. However, this active system has been amended to become fixed from both ends by Leemann et al. (2011) while Younis (2014) modified this approach focusing on restraining three prisms in one frame at the same time. The most well-known method is the ring test suggested by AASHTO T334-08 (Grzybowski and Shah 1990; Choi et al. 2015) but Loser and Leemann (2009) changed the geometrical arrangement of this test from circular to square to increase the number of cracks. In this system, the corners were fixed and the edges had reduced sections to pre-determine the crack location. Overall, different tests can simulate different applications but the best restrained method is the one closest to the real-life application(Banthia et al. 1996). However, adding fibres to concrete controls cracking behaviour in which smaller and narrower cracks can take place at the same time. Hence, a method which can accommodate several cracks is here suited to examine the behaviour of FRC.

Randomly distributed fibres in concrete can enhance tensile capacities and provide a better control of shrinkage cracking (Banthia et al. 1996; Younis 2014; Buratti et al. 2013). Hence, SFRC is widely used by the construction industry for various structural applications. The increase in the demand of FRC opens new aspects of research about management of the resources and the consumed energy in manufacturing such fibres (Pilakoutas et al. 2004; Groli 2014). Although manufactured fibres have been used for many decades as reinforcement in concrete, recently a new steel fibre (Recycled Tyre Steel Fibre – RTSF) has been developed as a by-product of mechanical shredding of end-of-life tyres. It has been found that RTSF can control concrete cracking by shrinkage due to bridging, but the overall shrinkage strains remain unchanged (Younis 2014; Graeff 2011).

Some research studies argue that shrinkage strains of FRC increase (compared to the plain concrete) due to the increase in the air voids while others claimed a decrease in the shrinkage strains due to the provided internal restraint by the fibres (Younis 2014; Graeff 2011). However, most of the published studies (Younis 2014; Jafarifar 2012; Graeff 2011) used a single type of fibres per mix and did not blend manufactured with recycled fibres. Hence, the effect of hybrid steel fibres (both manufactured and recycled) on concrete shrinkage should be investigated. To investigate the effect of hybrid steel fibres on the free and restrained concrete shrinkage, this study used the same methodology developed by Younis (2014).

2 Materials

2.1 Steel Fibres

Two types of steel fibres were used in this experimental programme, manufactured undulated (MUND) and recycled tyre steel fibres (RTSF) as shown in Fig. 1a. Two aspect ratios of MUND were used with length/diameter of (l/Ø) 55/0.8 and 60/1. The geometrical characteristics of RTSF vary due to the methodology used to retrieve them from the tyres. In this study, the average diameter was of 0.15 mm, the average length at 50% cumulative mass was about 20 mm, less than 10% of the fibres were less than 8 mm long and less than 10% were longer than 34 mm (see Fig. 1b).

Fig. 1.
figure 1

Types of steel fibres (a) and RTSF distribution (b).

Full size image

2.2 Concrete Mix Design

The concrete used in this experimental programme was delivered by a third party ready mix supplier and the concrete mix design is shown in Table 1 (though the mix design was the same, the workability differed since the batches were delivered in three different days). The mix design consisted of CEM 1 (50%) and GGBS (50%). GGBS was produced in accordance with BS EN 15167-1:2006.

Table 1. Mix ingredients.
Full size table

3 Experimental Programme

The mix programme consisted of seven SFRC mixes and one plain concrete mix for comparison purposes as shown in Table 2 (M stands for Manufactured fibres, R for Recycled and T for textile whilst the number next to each fibre type denotes the fibre content). These mixes were cast in three different batches and a plain concrete mix was cast for each batch. MUND and RTSF were used on their own or blended together as reinforcement in concrete to investigate the free and restrained shrinkage behaviour of SFRC. Three different fibre contents were investigated (30, 35 or 45 kg/m3) to optimise the best candidate. Mix M20R10T1 contained 1 kg/m3 of recycled textile obtained from the end-of-life tyres. Steel fibres were dispersed manually into the concrete and mixed for at least 5 min to achieve a uniform distribution. Each mix comprised 3 prisms (100 × 100 × 500 mm) for free shrinkage measurement and three prisms which were restrained in a steel frame as shown in Fig. 2 (Younis 2014). All specimens were stored under the same controlled environmental conditions (temp: 23 ± 2 °C and RH: 40 ± 5%). Shrinkage measurements were taken by a 200 mm “Demec” strain gauge.

Table 2. Experimental programme.
Full size table
Fig. 2.
figure 2

Restraining shrinkage schematics (a) steel frame model, (b) frame after casting.

Full size image

4 Results and Discussion

4.1 Free Shrinkage

The total free shrinkage was measured by using a 200 mm Demec gauge at top and bottom of both sides of the prisms. Then, the measurements were averaged and the results are shown in Fig. 3. In general, the SFRC mixes showed higher shrinkage strains than the plain concrete mixes. The discrepancies in shrinkage measurements between SFRC mixes may be attributed to the amount of initial water content in each mix as well as the air content. Hence, the effect of steel fibre type or dosage in shrinkage is difficult to be decoupled. When the shrinkage experimental measurements are compared to the model predictions of Euro Code (EC-1 2004) and fib Model Code (MC-2010), there is a significant overestimation. This may happen for the following reasons:

Fig. 3.
figure 3

Free shrinkage strains.

Full size image
  1. a.

    The binder type: The EC and fib models do not consider cement replacement materials for their predictions. The mixes of this study included 50% GGBS as well as 50% CEM1. In literature, GGBS was found to reduce the total shrinkage amount (Jianyong and Yan 2001; Aly and Sanjayan 2008; Mitchell and Arya 2015) as the fineness of GGBS seems to close the concrete pores which make water unable to escape from the substrate(Castel et al. 2016). Hence, shrinkage developed at a slower pace and giving lower values than the predictions of the design codes.

  2. b.

    Due to the high amount of Demec points bonded on the prisms, the first shrinkage measurement was concluded within 6 h; while the codes recommend to take the measurements within 3 min.

The small undulation in the curves is a result of the temperature and relative humidity change in the control room and the expansion at 240 days was due to accidental water spill on the floor in the control room.

4.2 Restrained Shrinkage

Figure 4 shows the average shrinkage measurements from the Demec gauge 200 mm at the top and bottom of both faces of each specimen. All mixes exhibited similar restrained shrinkage apart from mixes M35 and R30. The shrinkage of these two mixes started deviating from the rest of the mixes between the age of 14 and 28 days, where no significant development in shrinkage took place in mix M35 whilst there was a remarkable increase in mix R30. For the mixes with similar shrinkage performance, it may be concluded that steel fibres type and dosage did not affect restrained shrinkage; same as in free shrinkage. This shows that such small dosages of steel fibres do not provide significant internal restraint for concrete. Since restrained shrinkage specimens were stored in the same room as the free shrinkage specimens, a similar drop in shrinkage took place at around 240 days due to an accidental water spill (Fig. 3). An overall comparison of the free and restrained shrinkage strain (using this particular passive restraint steel frame arrangement) shows that the restrained shrinkage strains were about half the strain measured in the free shrinkage specimens.

Fig. 4.
figure 4

Restrained shrinkage strains.

Full size image

4.3 Restrained Factor

The Restrained Factor (RF) is calculated as \( RF = \frac{{\varepsilon_{sh,free} - \varepsilon_{sh,restrained} }}{{\varepsilon_{sh,free} }} \). On average, the restrained factor for all mixes (except mix R30) varied between 0.5 and 0.6 as shown in Fig. 5 (a slight increase in RF took place after day 240). Mix R30 exhibited very low RF because it showed the lowest free shrinkage development and the highest restrained shrinkage. It is observed from Figs. 3 and 4 that there is a reduction in the restrained shrinkage after 150 days while the free shrinkage remained the same. This may be attributed to stress relaxation in the anchors or creep.

Fig. 5.
figure 5

Restrained factor.

Full size image

5 Conclusions

From the results, the following conclusions can be drawn:

  • The shrinkage of SFRC was higher than the shrinkage of the plain concrete probably due to the increase in the air voids.

  • Free shrinkage was much lower than what was predicted by the design codes due to the use of GGBS which closed the pore holes and prevented water to evaporate.

  • Steel fibres did not provide any significant internal restraint for the concrete shrinkage.

  • External restraint reduced the shrinkage strains by about 50%.