Assessment of activity moduli and acidic resistance of slag-based geopolymer concrete incorporating pozzolan

The environmental impact of Portland cement production and utilization in the construction sector has led to the global call for the use of eco-friendly construction materials for the production of cleaner and sustainable products. Therefore, this study explored agro-industrial wastes, slag and corncob ash, for the production of geopolymer concrete (GPC). Corncob was dehydroxylated at 600 C for 3 h and partially used as a replacement for slag at 0%, 20 %, 40 %, 60 %, 80 %, and 100 %. A 12 M, 14 M, and 16 M of both sodium silicate (SS) and sodium hydroxide (SH) were used as activators. The chemical moduli of each and mixed binder were quantified and evaluated based on the major reactive oxides, hence leading to the evaluation of reactivity indexes (RIs). Moreover, the RIs and mix design properties (MDPs) of concrete were used for the prediction of flexural strength while the chemical resistance of each concrete sample was investigated. Compared with the experimental results, the predictive flexural strengths based on the RIs and the MDPs yielded a high precision with R ranging from 88–92 % at 7–90 days, respectively. Moreover, the GPC, unlike Portland cement concrete (PCC), resisted the more acidic attack. Therefore, the use of GGBFS CCA blended concrete would be more advantageous in a highly acidic environment than PCC. Ultimately, the models proposed by this study can be useful in the concrete mix design procedure for the flexural strength development of GPC incorporating agro-industrial provided the oxide compositions of each and mixed material were obtained. © 2020 The Authors. Published by Elsevier Ltd. All rights reserved.


Introduction
In the construction sector, the utilized rate of concrete is high, owing to the rapid industrialization and urbanization [1]. Amongst the concrete constituents, Portland cement (PC) plays a pivotal role in determining the quality of concrete. However, the production of PC, apart from its negative impact on the environment, requires a massive industrial process [2]. A ton of PC production, which emits 1 ton of carbon dioxide (CO 2 ) into the atmosphere, requires 4000 MJ energy, 1.5 tons of raw materials, and 140 kW h of electricity [3]. Moreover, a 7-9 % of CO 2 is emitted yearly into the atmosphere following the massive requirements of energy in PC production, hence contributing to the serious global warming [4]; this poses huge threats to human and ecosystem survival and development. Besides, from the building activities alone, Mahmoudkelaye et al. [5] estimated 30-40% generation of greenhouse gas (GHG) emissions, globally. Moreover, the yearly utilization of PCC in the construction industry is estimated to be 20 billion tons globally [6]. Furthermore, the need for the construction of construction cost and solid wastes, hence driving sustainability. The models developed from this study would enhance the findings of future studies on GPC incorporating SCMs by providing means of predicting fr based on RIs and MDPs.

Materials
The locally sourced materials, GGBFS and CCA, as shown in Fig. 1, were used as SCMs for the production of GPC, while Portland limestone cement (PLC), as shown in Fig. 1, was used as a binder for the production of PCC and compared with GPC. Slag was ground to obtain GGBFS. Corncob was dehydroxylated at 600 C for 3 h to obtain CCA. The SCMs were then sieved with BS 90 mm to obtain a similar fineness with PLC.
The specific gravity (SG) of the binding materials was determined following the requirements stated in BS EN 196À3 [44] using a specific gravity bottle and kerosene. The results indicated 2.90 g/cm 3 , 2.44 g/cm 3 , and 3.15 g/cm 3 for GGBFS, CCA, and PLC, respectively. Owing to these results, GGBFS met the SG limit of 2.90 g/cm 3 to 3.15 g/cm 3 specified by BS EN 15167À1 [38], while that of CCA confirmed the similar results obtained by Oyebisi et al. [21].
The fineness of binding materials was determined using the dry sieving method, and BS sieve 90 mm as stipulated by BS EN 196À6 [45]. The results showed 7.6 %, 8.0 %, and 7.5 % for GGBFS, CCA, and PLC, respectively, hence satisfying the 12 % maximum fineness specification prescribed by BS EN 196À6 [45]. Therefore, the materials are suitable for use as binder and SCMs in concrete production. Furthermore, Laser diffraction, Model Beckman Coulter LS-100, was used to analyze the particle size distribution of the binding materials, as shown in Fig. 2, over the range size of 0.5 mm-900 mm. The results indicated a mean particle size of 20.68 mm, 23.45 mm, and 18.79 mm for GGBFS, CCA, and PLC, respectively. Besides, the specific surface area was carried out on the binding materials following the procedure stated by BS EN 196À6 [45] using the Blaine method at a standard porosity of 0.500. The results indicated 420 m 2 /kg, 625 m 2 /kg, and 375 m 2 /kg for GGBFS, CCA, and PLC, respectively.
The alkaline solutions, SH pellets with 99 % purity, and SS gel were locally sourced and used as activators. SS gel comprises Na 2 O, SiO 2 , and H 2 O of 9.4 %, 30.1 %, and 60.5 % respectively, with SiO 2 /Na 2 O weight ratio of 3.20 and S.G. of 1.40 g/cm 3 at 20 C. A 354 g, 400 g, and 443 g of SH pellets were measured and dissolved in 646 g, 600 g, and 557 g of clean water based on the chemistry procedures established by Rajamane and Jeyalakshmi [46] for the preparation of 12 M, 14 M, and 16 M activators, respectively. The SH solutions were prepared 24 h earlier to reduce the high rise in temperature owing to the reaction between SH pellets and water and added to SS gel 2 h before casting for better performance, using a SS/SH ratio of 2.5: 1.
The locally sourced aggregates were used and prepared at saturated surface conditions before the mix design. Grading was also conducted on the aggregates to obtain the needed particle size distribution (PSD). Moreover, the aggregates were characterized in line with the BS EN 12,620 [47]. The specific gravity (SG), water absorption (WA) and moisture content (MC) of the aggregates were determined following the procedure stated in BS EN 12,620 [47]. The results showed the SG of 2.60 g/ cm 3 and 2.64 g/cm 3 ; WA of 0.7 % and 0.8 %, and MC of 0.3 % and 0.2 % for both fine aggregate (FA) and coarse aggregate (CA), respectively. Fig. 3 shows the PSD of both FA and CA used; the aggregates satisfied the limits of BS EN 12,620 [47]. On the other hand, the mineralogical composition of the coarse aggregate (granite) was identified with the aid of the Petrological Microscope, Model RPI-3 T. The sample was prepared, polished in a glass ground plate using a carborundum, and mounted on a clean glass slide with adhesive [48]. Also, the chemical composition was analyzed using the XRF spectrometer machine, Philips PW-1800. The results of mineralogical composition showed quartz, feldspar, mica, and iron oxide of 62  ranged between 1.2-3.5 for calcalkalinity [35]. XRF analysis was not performed on the FA because it comprises SiO 2 content, almost in its entirety [48].

Strength activity index (SAI)
The SAI was determined in line with the BS 3892À1 [40]. The water to binder ratio was modified to allow for the same flow properties with reference-cement mortar cubes [49], hence requiring mixing water of 235 mL of distilled water. Therefore, the SAI was determined for both 7 days and 28 days compressive strengths (CS) on the average of three samples using the relationship, as illustrated in Eq. 1 [40].
where P is the average CS of pozzolan-reference cement mortar cubes (in MPa) C is the average CS of reference-cement mortar cubes (in MPa). The cement-reference mortar cubes, at 7 days and 28 days, showed the CS of 40.64 MPa and 50.43 MPa, respectively. Following the formula, as illustrated in Eq. 1, the test pozzolan (CCA) exhibited a SAI of 0.85 and 0.91 at 7 days and 28 days, respectively, thus showing considerable pozzolanic activity because SAI is greater than 0.80, as recommended by BS 3892À1 [40]. The cumulative particle size distribution of binding materials used. LL is the lower limit, UL is an upper limit, FA is fine aggregate, and CA is coarse aggregate

Frattini test (FT)
The FT was analyzed following the procedure specified by BS EN 195À5 [50]. The theoretical maximum concentration (TMC) of [CaO] was determined using the relationship, as illustrated in Eq. 2 [50]. As shown in Table 1

. Slag activity index
The method was carried out following the procedure in ASTM C 989 [39] for the GGBFS. The procedure is similar to that of SAI stated in BS 3892À1 [40] except for 50 % cement replacement, C 3 A content ranging from 6 to 10 %, and a maximum of 3% SO 3 content specified by the standards for the reference cement. Therefore, from the XRF results, PLC exhibits 2.03 % SO 3 content, hence satisfying the maximum requirement of 3%. Besides, C 3 A was quantified based on Bogue's equation, as shown in Eq. 3 [34]. Based on the XRF result and in line with Eq. 3, the result exhibited 10 % C 3 A, thus fulfilling the maximum specification of 10 %. Finally, the slag activity index was determined following the relationship, as illustrated in Eq. 1.
The CS of GGBFS-reference cement mortar cubes to the cement-reference mortar cubes with the mean particle size (d v 50 = 20.68 mm) exhibited the slag activity index of 76.42 % and 98.53 % at 7 days and 28 days, respectively, hence classifying as grade 100 because the activity index is more than 70 % and 90 % at 7 days and 28 days, respectively [39].

Caustic soda test (CST)
This method was carried out following the procedure outlined in ASTM C 1073À18 [42]. The diluted solution-to-GGBFS ratio was fixed at 0.5 and used to prepare 40 mm Â 40 mm Â 160 mm prismatic samples. After

Materials characterization
The oxide compositions of binding materials, CCA, GGBFS, and PLC, were analyzed using the XRF spectrophotometer machine, Philips PW-1800. The results are shown in Fig. 4. The results revealed that CCA satisfied the chemical pozzolanic requirements stipulated by BS EN 450À1 [51] and BS EN 8615À2 [52] such that the addition of SiO 2 , Al 2 O 3 , and Fe 2 O 3 met 70 % minimum requirement. The content of CaO within the range of 10-20% established by Al-Akhras [53] was also met. It can be deduced that the CCA could exhibit a pozzolanic reaction and used as the SCM in the production of blended GPC. On the other hand, GGBFS met the BS EN 15167À1 [38]'s limit requirements of 32-40% for both silica (SiO 2 ) and lime (CaO) contents. Besides, (CaO + MgO/SiO 2 ) ! 1, (CaO/SiO 2 ) 1.4, and SiO 2 + CaO + MgO ! 67 % stipulated by BS EN 15167À1 [38] were also met. Also, the oxide compositions obtained herein for GGBFS showed similar compositions with the previous studies [19,31]. Therefore, an inference is made that GGBFS utilized in this study could exhibit both pozzolanic and selfcementitious reactivity, hence suitable for use. In the same vein, the PLC fulfilled the chemical requirements specified by BS EN 196À2 [54].
The microstructural behaviour of the binding materials, GGBFS, CCA, and PLC, was examined using the SEM machine, JEOL 7000600, to establish the characteristics that influenced the RIs of each binder. The SEM analysis was performed on a flat (general) scan. For the investigation, the accelerated voltage was constant at 15 kV, while images were observed at 4000x magnification in a high vacuum. The SEM micrograph results are presented in Fig. 5(a), to a limited extent, reveals a wrinkled internal structure with sharp needles. However, Fig. 5(b) shows an amorphous structure, while Fig. 5 (c) reveals a crystalline and spherical structure.

Mix design quantities
The mix quantities were designed following the procedures stated by BS EN 206 [55]. The percentage replacement of GGBFS by CCA was selected to examine the replacement levels, which would meet the target strengths for both structural and non-load bearing applications. Owing to this, GGBFS was replaced with CCA at 0%, 20 %, 40 %, 60 %, 80 %, and 100 % for the production of GPC and was respectively indicated as E1, E2, E3, E4, E5, and E6, while the PCC (100 % PLC) was indicated as E0.  Tables 2 and 3, respectively.

Mix preparation, casting and curing
The dry constituents were prepared following the procedures prescribed by BS 1881À125 [56] and BS EN 12390À2 [57] by preparing and pouring fresh concrete into a cubical mould of 150 mm 3 for compressive strength test, and 150 mm Â 600 mm long beam for flexural strength test. The fresh sample was randomly compacted each in three layers, cured under 25 C and 65 % RH. The compressive strength was tested at 90 days curing, while the flexural strength was tested at 7, 28, 56, and 90 days.  The compressive strength (f c ) and flexural strength (f r ) were determined with the aid of an INSTRON 5000R UTM following the procedures stated by BS EN 12390À4 [58] and BS EN 12390À5 [59] in a constant force regime under a loading rate of 0.6 MPa and 0.06 MPa per second, respectively. Three (3) samples were made and crushed for each mix ID, and the average was used for the analysis.

Reactivity indexes (RIs) of binding materials
The RIs of binding materials were evaluated using the principal reactive oxides such as CaO, SiO 2 , Al 2 O 3 , Fe 2 O 3 , MgO, and SO 3 following the establishment of their oxide compositions, which reflect both self-cementitious and pozzolanic reactivity [14,18,32]. The concept which guides the RIs is illustrated in Eq. 4-8 as reactivity, hydraulic, lime, silica, and alumina moduli of each and blended binder, indicating as RM, HM, LM, SM, and AM, respectively.  w/b is the water to binder ratio = 0.54; b/agg is the binder to aggregate ratio = 0.23.

Prediction of f r based on RIs and MDPs
For this study, MDPs are indicated as water to binder (w/b) ratio and binder to aggregate (b/agg) ratio. Either reactivity, hydraulic, or lime moduli quantifies the self-cementitious properties of each and blended binding material while the pozzolanic activity is quantified by both silica and alumina moduli [32,37]. Consequently, a linear relationship exists in the prediction of f r and RIs. Thus, the regression was first modelled based on the combination of RM, SM, and AM; HM, SM, and AM; and LM, SM, and AM using Minitab 17 statistical software. Furthermore, in determining the f r of blended concrete, the RIs of blended binders were integrated and normalized with an inverse of w/b ratio; hence, f r becomes a direct proportion to RIs, but an inverse proportion to w/b ratio [14,18]. Therefore, the fit regression relationship between fc and w/b ratio was first normalized and modelled in the range of 0.54 to 0.42 w/b ratio for grades M 30 to M 40 concrete, respectively. The f r and RIs were selected as the response (dependent variable) and continuous predictors (independent variables), respectively, to predict the design data in Minitab 17.
The binder to aggregate (b/agg) ratio also contributed a vital role to the evaluation and improvement of the concrete strength apart from RIs and w/b ratio [14]. The f r of blended binders was significantly improved when RIs, w/b ratio, and b/agg were all used for the strength correlation. It is noteworthy to state that the volume ratio was used to model the b/agg ratio against the weight ratio. For each mix, the volume fraction was determined using its moisture content and specific gravity to improve the binder-aggregate packing capacity. Following the incorporation of w/b ratio, the fit regression relationship between f r and b/agg ratio was modelled in the range of 0.31 to 0.23 b/agg ratio for grades M 30 to M 40 concrete, respectively. Consequently, f r was predicted based on the RIs and MDPs, as illustrated in Eq. 9-11.

Durability test
The chemical resistance was conducted on the prepared cube samples using the solutions of sulphuric acid (H 2 SO 4 ) at 2% concentration [34,60] for acidic attacks. The concrete specimens were tested for both weight and strength loss after 90 days of immersion in H 2 SO 4 .  The results could be associated with the reaction between the aluminosilicate glassy phases (amorphous structure) of GGBFS, as shown in Fig. 5(b), and the alkaline solutions, hence resulting in x-ray amorphous aluminosilicate paste (X-RAAP). The X-RAAP, according to Criado et al. [61], contributes to the higher strengths of the hardened product, compared with both wrinkled and crystalline structures for PCC and CCA, as shown in Fig. 5(a) and (c), respectively. Besides, unlike 12 M and 16 M activators, 14 M activator exhibited the highest compressive and flexural strengths at all curing days because of its capacity to liberate more aluminosilicate gels in the mix. However, at 16 M activating level, the OH À solution in the mix could be excessive, thus encasing the amorphous paste, causing a barrier to the activating dissolution, and delaying the hydrating agent (calcium-silicate-aluminate-hydrate, C-S-A-H) in the mixed paste; this delays and decreases the strengths.   Fig. 8 shows a decrease in CaO, Al 2 O 3 , and MgO with increasing CCA content, while SiO 2 , Fe 2 O 3 , and SO 3 increase with increasing CCA content in the blended mix; this supports the similar findings reported by Akinwumi and Aidomoje [17] that the reactive oxides, CaO, MgO, and Al 2 O 3 decrease with increasing CCA content, while SiO 2 , Fe 2 O 3 , and SO 3 increase with increasing CCA content for the CCA-PC blend. Meanwhile, Behim et al. [32] and Demoulian et al. [36] stated that GGBFS exhibits a similar mineralogical composition to PC; it majorly possesses oxides of Ca, Si, Al, Mg, and Fe, and this gives GGBFS its hydraulic and pozzolanic properties. Also, Xia and Visintin [14] and Darquennes [33] opined that slag is said to exhibit both self-cementitious and pozzolanic properties if the content of CaO and SiO 2 is higher than 30 %. From the XRF results of GGBFS, it is clear that the contents of both CaO and SiO 2 are higher than 30 %. On the other hand, Taylor [62] and Hewlett [63] reported that the self-cementitious reaction of slag decreases as the crystalline content in the blended mix increases; this demonstrates that the reactivity of GGBFS depends on the increasing content of its amorphous structure, and the significant oxides which contribute to the high phase of an amorphous structure are oxides of Ca, Al, and Mg [64,65]. Thus, through close examination of microstructures of binding materials, as shown in Fig. 5, it was evident that the content of the amorphous structure in GGBFS could gradually decrease while the content of the crystalline structure in CCA might increase when GGBFS is replaced with CCA. Consequently, as the content of CCA in the blended mix increases, CaO, Al 2 O 3 , and MgO in GGBFS decrease. In contrast, SiO 2 , Fe 2 O 3 , and SO 3 in CCA increase; this corroborates the findings from relevant studies in that the reactivity of GGBFS increases with increasing CaO, MgO, and Al 2 O 3 contents but reduces as the contents of SiO 2 , Fe 2 O 3 , and SO 3 rise [66,67]. However, it was pointed out that GGBFS comprises small crystal material and is advantageous to its reactivity [68][69][70]. Besides, Gruskovanjak et al. [67] pointed out that the optimum content of the principal reactive oxides of slag is more beneficial to its self-cementitious reactivity than the content of the amorphous structure. Therefore, it is inferred that the contents of CaO, Al 2 O 3 , MgO, SiO 2 , Fe 2 O 3 , and SO 3 influence the reactive potentials of GGBFSÀCCA blended binders.

RIs of the blended mix
In assessing the RIs of each blended binder, Eq. 4-5 was used, and the results are shown in Fig. 9. It was revealed that the RM, HM, LM, and AM decreased with increasing CCA content, while the SM increased with increasing CCA content in the blended mix for both M 30 and M 40. Besides, it was evident from the results that CaO, Al 2 O 3 , MgO, SiO 2 , Fe 2 O 3 , and SO 3 influenced the RIs of the blended binders. The RM, HM, and LM of the blended binders increased with increasing CaO, Al 2 O 3 , MgO contents, while the SM and AM of the blended binders increased as the contents of SiO 2 and Al 2 O 3 increased, respectively. In contrast to HM, the RM of the blended binders met the minimum requirement of 1.0 specified by BS EN 8615À2 [52]. Statistically, the RM, HM, LM, and AM of the blended mix decreased from 25 to 78 %, 19-77%, 19-77%, and 11-26% as the percentage replacement of CCA by GGBFS increased from 20 % to 100 % for both M 30 and M 40, respectively. The self-cementitious properties of mixed binders increase with an increase in CaO, Al 2 O 3 , and MgO contents, thus resulting in stronger hydraulic reactions [32,71]. As a result, the decrease in RIs may be attributed to the reduction in principal reactive oxides, CaO, Al 2 O 3 , and MgO as a result of the increase in CCA content in the blended mix. However, the SM of the blended mix decreased from 44 to 10 % as the percentage replacement of CCA by GGBFS rose from 20 to 100 % for both M 30 and M 40, respectively. Meanwhile, it is shown in Fig. 7 that CCA, being a pozzolan, exhibits higher content of silica (SiO 2 ) compared with that of GGBFS. Therefore, this result confirms the findings reported by Mathhes et al. [70] that SM increases with increasing SiO 2 content, hence resulting in stronger pozzolanic properties. On the other hand, the reactivity of GGBFS depends on its amorphous structure, thus influencing its RIs [69,70]. This assertion confirms the SEM micrographs, as shown in Fig. 5 (b) and (c) for GGBFS and CCA, which display amorphous and crystalline structures, respectively.

Prediction of f r based on RM, AM, SM, and MDPs
Following Eq. 9, the results of the statistical data are shown in Fig. 10 (a)-(d) for 7, 28, 56, and 90 days, respectively. It was observed that some data points for SM significantly deviated from the regression line. This may be asserted to the diversity of chemical compositions of blended binders, aggregate type and volume, and mix design proportions; this assertion confirms the findings reported by Xie and Visintin [21] and Neville [34] that differences in the oxide composition of blended binders, aggregate types, texture, and shape, and methods of mix design, affect the data results, hence influencing the reactive potentials of blended concrete incorporating SCMs. Moreover, the flexural strength increased with increasing RM and AM but decreasing SM; this may be attributed to the higher contents of CaO and Al 2 O 3 in GGBFS, which increases RM and AM, thus resulting in a stronger self-cementitious reaction. However, the higher content of SiO 2 in CCA increases its SM, hence leading to a pozzolanic reaction rather than a hydraulic reaction; this also confirms the findings of a similar study reported by Gruskovnjak et al. [67] that the RM and AM increase as the CaO and Al 2 O 3 contents increase, while SiO 2 content reduces, thus resulting in high reactivity. However, the higher contents of SiO 2 and low contents of CaO and Al 2 O 3 result in low reactivity. On the other hand, a blended mix with high contents of CaO, Al 2 O 3 , and MgO exhibits high self-cementitious/hydraulic properties in the presence of alkaline activators [66,68,70].
The fit regression model was used for the correlation of f r based on the RIs (RM, AM, and SM) and MDPs at the global trend of 95 % confidence interval (CI) and prediction interval (PI). Thus, the regression equations are illustrated in Eq. 12-15 for 7, 28, 56, and 90 days, respectively. Therefore, the coefficient of determination (R 2 ) is 87.47 %, 87.60 %, 88.12 %, and 92.26 % fit to predict the data at 95 % CI and PI for 7, 28, 56, and 90 days, respectively, thus indicating 0-5.20% increase in R 2 as the curing age increases from 7-90 days. Ultimately, relative to RIs and MDPs, these developed models can be used for the strength prediction of concrete incorporating SCMs. , and 90 days, respectively. It was observed that some data points of SM were out of the regression line due to the difference in binders' oxide compositions, the volume and chemical compositions of aggregates, and mix proportions. Besides, the f r of GGBFSÀCCA blended concrete increased with increasing HM and AM but decreasing SM. The reason for a higher strength cannot be far-fetched: GGBFS exhibits higher content of CaO and Al 2 O 3 compared with CCA, hence resulting in stronger hydraulic reaction, but this hydraulic reaction decreases when replaced with CCA, which predominantly contains a higher content of SiO 2 . This assertion is in line with the findings reported in various studies that the hydraulic response of slag reduces with increasing silica content [64,65]. Therefore, it is inferred that the HM of GGBFSÀCCA blended binder increases with higher contents of CaO and Al 2 O 3 and the lower content of SiO 2 in the mix. The fit regression model was used for the correlation of f r based on the RIs (HM, AM, and SM) and MDPs at the 95 % CI and PI, and the regression equations are illustrated in Eq. 16

Prediction of f r based on LM, AM, SM, and MDPs
The statistical data for LM, AM, SM, and MDPs are indicated in Fig. 12 (a)-(d) for 7, 28, 56, and 90 days, respectively. It was noticed that some data points of SM were out of the global trend due to the diversity in oxide compositions, aggregates volumes and types, and mix proportions of the blended binders. Moreover, the f r of GGBFSÀCCA blended concrete increased with increasing LM and AM but decreasing SM; this may be attributed to the fact that GGBFS exhibits higher content of CaO and Al 2 O 3 compared with CCA, hence resulting in a stronger reactive component. Still, this reactive component decreases when replaced with CCA, which majorly contains a higher content of silica. Therefore, it is inferred that the LM of GGBFSÀCCA blended binder increases with higher contents of CaO and Al 2 O 3 and the lower content of SiO 2 in the mix.
The f r , RIs (LM, AM, and SM), and MDPs were predicted using the fit regression model at the 95 % CI and PI, and the regression equations are illustrated in Eq. 20-23 for 7, 28, 56, and 90 days, respectively. Therefore, at 7, 28, 56, and 90 days, R 2 is 87.50 %, 87.65 %, 88.09 %, and 92.25 % fit to correlate data, respectively, hence indicating 0-5.20% increase in R 2 as the curing age increases from 7 to 90 days. Finally, with respect to RIs and MDPs, these proposed models can be used for the strength prediction of concrete incorporating SCMs.  Fig. 13 illustrates the statistical comparison and trend between the flexural strengths of experimental results and that of predictive values. It was observed that both empirical and predictive results exhibited similar values and patterns of flexural strength. In contrast to HM, both LM and RM showed the best fit at all levels of curing time for both M 30 and M 40. These observations confirm the findings reported in similar studies such that LM yields the best fit for PC blended with cashew nut shell ash (CNSA) [18]. In contrast, RM yields the best fit for blended concrete incorporating SCMs [14]. Despite producing the  Fig. 9 that LM of the GGBFSÀCCA blended binders was low compared with the minimum requirements (! 0.66 1.02) recommended by BS EN 197À1 [72]; besides, HM was less than 1 compared with the minimum requirement (< 1) specified by BS EN 197À1 [72]. However, RM satisfied the minimum requirement (< 1) defined by Behim et al. [32], Demoulian et al. [36], and BS EN 15167À1 [38]. The variations in LM and HM may be attributed to the difference in chemical and mix properties of concrete in that BS EN 197À1 [72]'s recommendation was based on the PC blended binders such that the ratio of CaO to SiO 2 in the blended mix was high compared with GGBFSÀCCA blended binders reported in this study. Therefore, it is inferred that RM yields the best fit for GGBFSÀCCA blended binder, and this can be used in the validation of blended binders incorporating SCMs.

Conclusion
The study examined the GGBFSÀCCA-based GPC, and its effects on the activity indexes and the acidic attacks were evaluated. Both experimental and statistical methods were used in the course of the study, and the results were compared with PCC. Consequent upon the findings and in line with research aims, the following sets of conclusions are made: The reactivity of GGBFSÀCCA blended binder increases with increasing CaO, MgO, and Al 2 O 3 contents. However, the reactivity decreases with increasing SiO 2 , Fe 2 O 3 , and SO 3 contents. The RM, HM, and LM of GGBFSÀCCA blended binder increases with increasing CaO, MgO, and Al 2 O 3 contents, while the SM and AM increase with increasing SiO 2 and Al 2 O 3 contents. Flexural strength of GGBFSÀCCA GPC increases with increasing RM, HM, LM, and AM RM yields the best fit for predicting the flexural strength of slag-based GPC, incorporating CCA compared with HM and LM. Besides, a good correlation exists between the experimental results and proposed model equations.
There is a remarkable improvement in R 2 as the curing age increases. Slag-based GPC incorporating CCA provides excellent acidic resistance superior to that of PCC.
The concept of activity moduli in predicting the f r of the GGBFSÀCCA blended mix is attainable. This study benefits future research by focusing on three prospective solutions. First, the proposed model equations can be useful in the prediction and application of strength design proportions for GPC incorporating agro-industrial by-products under ambient curing conditions provided the chemical compositions are obtained. Second, the efficiency of the fit regression model in predicting f r based on the RIs and the MDPs is affirmed. Third, the application of agro-industrial by-products, GGBFS and CCA, can be advantageous in a highly acidic environment.

Declaration of Competing Interest
The authors show a credit to the sources in the manuscript. The authors declare that they have no known competing for financial interests or personal relationships that could have appeared to influence the work reported in this paper. The raw/ processed data required to reproduce these findings cannot be shared at this time as the Data also forms part of an ongoing study. The authors declare that the manuscript is the authors' original work and has not been published before. The authors also declare that the article contains no libellous or unlawful statements and does not infringe on the rights of others.