*3.2. Comparison of the Water Absorption and Apparent Porosity of Control Concrete and Binary and Ternary Concrete Mixtures*

The comparison of the other important properties (WA and AP) of the binary and ternary concrete mixtures with that of the CC was also performed in addition to the strength characteristics. This is because the durability of the hardened concretes can be indirectly assessed based on the WA and AP values. The different trends of changes in the WA and AP according to various amounts of cement substitution with WSA alone (binary concretes) and WSA with SF (ternary concretes) are depicted in Figures 3 and 4. It can be seen in Figure 3 that a lower WA compared to CC was exhibited by all binary and ternary concretes regardless of the amount of cement substitution, which is attributed to their lower AP values as compared to CC (Figure 4). A decrease in the WA of binary concrete mixtures with increasing percent substitution of cement with WSA was observed up to a certain replacement level (20%). Thus, a slightly higher WA was exhibited by binary concrete having 30% WSA when compared with other binary concretes having 10% or 20% WSA (Figure 3). These binary concretes also demonstrated a similar trend of AP (Figure 4). A slightly higher value of AP exhibited by concrete containing a high amount of WSA (30%) is probably due to a relatively slower rate of the pozzolanic reaction owing to replacement of high percentage of cement with WSA. On the contrary, ternary concretes having high cement substitutions with blends of WSA and SF demonstrated a remarkable decrease in WA and AP values. The results demonstrated that, despite similar cement substitutions (30%), the ternary mix WSA25SF5 exhibited slightly lower WA and AP to that of corresponding binary mix WSA30. However, contrary to this, the WA and AP of this ternary mix were slightly higher than the other binary mixes with 10% and 20% WSA. Increasing the cement replacement from 30% to 40% in ternary concrete (WSA33SF7) resulted in further decrease in WA and AP values. The results demonstrated that the ternary concrete WSA33SF7 with an even higher percent replacement of cement exhibited lower WA and AP than the ternary concrete WSA25SF5 and all binary concretes having relatively lower cement substitutions (10%, 20%, or 30%). This was because SF particles of very fine size leads to significant pore refinement in ternary concrete mixtures compared to WSA. The ternary concrete having a very high cement substitution of 50% (WS40SF10) yielded slightly higher WA and AP values when compared to those of WSA33SF7 with relatively low cement substitution (40%). This was due to potentially slower rates of pore refinement and pozzolanic reaction in WSA40SF10 concrete owing to high cement substitutions.

**Figure 3.** Comparison of the results for water absorption between the control and other concretes (binary and ternary mixes) after 91 days of standard curing.

**Figure 4.** Comparison of the results for apparent porosity between control and other concretes (binary and ternary mixes) after 91 days of standard curing.

#### *3.3. Evaluation of Compressive and Tensile Strength Correlation of Concrete by Prediction Models*

Regardless of the type of mixture proportions, curing conditions, aging, or binder type and its content, the values of the experimental STS of concrete can be correlated with their corresponding compressive strength, mainly due to the existing consistency between their general trends of development with aging. In addition to the existing codes such as ACI 318 [51], ACI 363 [52], and CEB-FIP model code 1990 [53], various correlations between compressive and tensile strength were developed by researchers depending upon their specific experimental data of the curing and testing conditions, geometry of specimen, and the types of concrete [54–60]. Nevertheless, a consistent equation in general form [*fsp* = *a* (*f <sup>c</sup>* ) *<sup>b</sup>*] is used for this correlation by all researchers, including the existing model codes. This equation presents *fsp* as the unknown STS of concrete to be predicted (MPa), whereas the compressive strength obtained directly from the experiments is represented by *f <sup>c</sup>* (MPa). In the equation, the parameters a and b are the constants that consider the dissimilarity of increasing rate between both mechanical properties. Based on different test results by researchers, the value of b varies, for example, as 0.67, 0.50, and 0.71 by the CEB-FIP model code [53], ACI 318 [51], and Kim et al. [57], respectively. The reason for dissimilarity in the b values arises since both ACI 318 and Kim et al. used specified and mean compressive strength, respectively, while the CEB-FIP model code used compressive strength associated with the specific characteristic compressive strength. Moreover, experimental results of 28 days were used in developing most of these correlations using Type-I cement for normal concrete subjected to standard moist curing at 20 ◦C. Despite consideration of the effect of various influencing factors (different binder types, curing, and aging) on the rate of both properties by some researchers [57], the effects of some other influencing factors such as the type of concrete using SCMs, geometry of specimen, and seasonal variations were overlooked. Having considered this important factor, it is desirable to evaluate suitability of existing correlations to predict the STS of concretes produced in this study containing various percentage of WSA alone (WSA10, WSA20, WSA30) and blends of WSA with SF (WSA25SF5, WSA33SF7, WSA40SF10).

Figure 5 depicts various existing correlations between compressive and tensile strengths of concrete. To evaluate the STS based on the experimental compressive strength, the current results of various binary and ternary concrete specimens along with CC were drawn

with respect to 3, 7, and 28 days of aging and compared to existing models [51–60]. The experimental results of compressive strength were used to estimate the values of STS for the prediction models. With the exception of Noguchi-Tomosawa [58] and JSCE-2012 design codes [59], and De Larrard and Malier [60], all other existing models significantly overestimated STS, as shown in Figure 5. For the CC and WSA33SF7 at any known value of the compressive strength, a close match of the STS with that of the experimental compressive–tensile strength was predicted, regardless of aging, when using the JSCE-2012 model [*fsp* = 0.23 (*f <sup>c</sup>* ) 2/3]. The Noguchi–Tomosawa model [*fsp* = 0.291 (*f <sup>c</sup>* ) 0.637], on the other hand, resulted in a close match for the binary (WSA10, WSA20, WSA30) and other ternary (WSA25SF5, WSA40SF10) concretes. Similarly, a reasonably good estimate of the STS for any value of the compressive strength for these mixes was also predicted by the proposed model of De Larrard and Malier [*fsp* = 0.6 + 0.06 (*f <sup>c</sup>* )].

**Figure 5.** Comparison of the correlation between experimental compressive and tensile strengths for different concrete mixtures with existing prediction models.

From these findings, it is revealed that the correlation between the compressive and splitting tensile strengths of concrete is not significantly influenced by the type of binder and aging. Kim et al. [57] also noted the same independency with respect to the cement type, aging, and curing temperature with no effect on the correlation of compressive and tensile strengths of concrete. The models proposed by either De Larrard and Malier [60] or Noguchi-Tomosawa [58], hence, are considered safe in estimating the STS of all the concrete mixtures studied with the exception of the CC and WSA33SF7 concrete. The JSCE model [59], however, accurately predicted STS of the CC and WSA33SF7 concrete. Furthermore, the JSCE model, with slight underestimation, satisfactorily predicted the STS of all the concrete tested. The underestimation of the STS with respect to the current experimental values is considered safe because it is used as the criterion of crack control.

#### *3.4. SEM–EDS Analysis of Control (C), Binary (C/WSA), and Ternary (C/WSA/SF) Cementitious Pastes*

Figure 6 shows the results of SEM–EDS analysis for different mixes. As shown in this figure, the effects of WSA and SF on the microstructure of cementitious paste were examined through EDS analyses that were performed on SEM micrographs. The purpose of the EDS analysis with SEM was to study the crystal structure changes in C-H and C-S-H phases of paste matrix. Based on the computation from EDS analyses, a comparison of Ca/Si ratios among different mixes is presented in Table 6.

**Figure 6.** *Cont*.

**Figure 6.** *Cont*.

**Figure 6.** SEM–EDS spectrum of control, binary (C/WSA), and ternary (C/WSA/SF) paste samples (**a**,**b**) control, (**c**,**d**) WSA10, (**e**,**f**) WSA20, (**g**,**h**) WSA30, (**i**,**j**) WSA25SF5, (**k**,**l**) WSA33SF7, (**m**,**n**) WSA40SF10.

**Table 6.** Ca/Si atomic ratio from EDS analysis of control and mixes containing different percentages of WSA alone and WSA with SF.


According to the findings of past researchers [61–63], the Ca/Si ratio for the C-S-H phase in paste sample ranges between 0.5 to 2.0, while 2.0 or higher for C-H phases. Among all mixes, both C-H and C-S-H phases of the control mix exhibited a highest Ca/Si ratio, at 3.30 and 1.93, respectively. As illustrated in Table 6, the lower Ca/Si ratio values of both the binary and ternary mixes is due to the substitution of cement with WSA and SF that led to decreased porosity of the paste matrix by forming high-density C-H and C-S-H phases in these mixes. Among binary mixes, WSA20 exhibited lowest Ca/Si ratio at 2.65 and 1.42 for C-H and C-S-H, respectively. Therefore, current results suggest a 20% replacement of cement with WSA without compromising the mechanical and microstructural performance of concrete. Moreover, all the ternary mixes showed even lower Ca/Si ratio as compared to binary mixes. A significant decrease in Ca/Si ratio in ternary mixes is attributed mainly to the fine, amorphous, and highly reactive nature of SF, which is used in ternary mixes in the presence of WSA. The addition of SF along with WSA results in the formation of additional C-S-H phases by utilizing C-H phases in the paste matrix. Furthermore, the addition of SF further enhances both the density of C-S-H and C-H phases and the compactness of cement paste. Among ternary mixes, the lowest Ca/Si ratio was observed for WSA33SF7, which indicates its improved microstructural properties due to the formation of high-density C-H (2.1) and C-S-H (0.91) phases. It is expected that such formation of high-density C-S-H and C-H phases would result in densification and refinement of the microstructure, leading to enhanced performance in practical engineering applications.
