Durability and mechanical properties of concretes with limestone filler with particle packing

2.1. Cement

The high-early strength Portland cement used followed the specifications from the Brazilian standard (29Associação Brasileira de Normas Técnicas. 2018. ABNT NBR 16697: Cimento Portland – Requisitos. ABNT, Rio de Janeiro.). This cement does not have any pozzolanic addition, and thus, the influence of this kind of mineral addition could be neglected. Table 1 summarizes the chemical composition of the cement used, obtained from X-ray fluorescence (FRX) test.

The specific mass of the cement was determined to be 3.04 g/cm³, according to Brazilian standard (30Associação Brasileira de Normas Técnicas. 2017. NBR 16605: Cimento Portland e outros materiais em pó – determinação da massa específica. ABNT, Rio de Janeiro.). In addition, a laser granulometry test was conducted on the material using laser diffraction in a CILAS 1090 SECO equipment, which can measure particles within a range of 0.10–500.00 μm, based on the theories of Mie and Fraunhofer, which is divided into 100 classes. The results of the gulometric analysis of the cement can be seen in Figure 1.

  

Figure 1 Laser granulometry of Portland cement and limestone filler. D10, D50 and D90 are the mesh sizes of the sieve through which 10%, 50% and 90% of the sample passes, respectively. 

2.2. Limestone filler

Limestone filler granulometry was performed using the same test and equipment described in section 2.1. From this test, the average size of the filler particles was 14.90 μm, as shown in Figure 2. Notably, the fineness of the filler was close to that of Portland cement, suggesting the possibility of partially replacing the binder by LF, as proposed by (31GraziaMT, SanchezLFM, RomanoRCO, PileggiRG. 2019. Investigation of the use of continuous particle packing models (PPMs). On the fresh and hardened properties of low-cement concrete (LCC). systems. Construct. Build. Mater.195: 524–536. 10.1016/j.conbuildmat.2018.11.051.) based on previous tests.

  

Figure 2 Aggregate granulometric curves. 

In addition, the chemical composition of the filler was determined through an X-ray fluorescence (FRX) test. The FRX results are summarized in Table 2.

2.3. Aggregates

For an appropriate packing, it is desirable to sort aggregates based on their granulometric characteristics. In the present case, sand was sorted and classified as fine, medium, and coarse sand. This sorting was achieved by sieving the sand thorough different meshes of sieves. To separate the coarse sand, a 1.2 mm sieve was chosen. Then, medium and fine sands were separated using a 0.6 mm sieve mesh.

As a result, sand 1 (coarse sand) was comprises the fraction of the sand retained in the 1.2 mm sieve, and the sand that passed through the 1.2 mm sieve and was retained in the 0.6 mm sieve was designated as sand 2 (medium sand). Finally, the sand that passed through the 0.6 mm sieve was designated as sand 3 (fine sand). To determine the granulometric composition of each sand, the specifications of the Brazilian standard (32Associação Brasileira de Normas Técnicas. 2022. NBR 17054: Agregados – determinação da composição granulométrica. ABNT, Rio de Janeiro.) were followed. Two different granite gravels were used in this work, gravel 0 (9.5 mm) and gravel 1 (19 mm).

The specific masses of the sands were obtained according to the Brazilian standard (33Associação Brasileira de Normas Técnicas. 2021. ABNT NBR 16916 Agregado miúdo - Determinação da densidade e da absorção de água. ABNT, Rio de Janeiro.).

Regarding that aggregates characterization is essential to achieve the particle packing objective, Figure 2 shows the granulometric curves of the different aggregates used in concrete mixtures.

Furthermore, Table 3 shows the specific masses of the used aggregates, which values with the granulometric distributions are helpful to build the concrete mixtures based on packing principles.

2.4. Concrete mixtures

2.4.1. Particle packing

 ⌅

To reach the appropriate particle size distribution and meet the packing condition, the modified A&A model (or Alfred) theoretical model was used. The virtual granulometric distribution should be approximately equal to the real system, as proven via numerical simulations and using real mixtures by (28SantosRA, MeiraGR, BezerraWVDC, BragaFAV, PontesDL. 2021. Use of numerical method for optimization of granulometric curves in eco-efficient concrete. Matéria, Rio de Janeiro. 26(4):e13115. 10.1590/S1517-707620210004.1315.).

To reach the suitable choice of the fraction of each aggregate, regarding the appropriate diameters of the particulate materials and the principle that the larger voids should be filled by the smaller particles, the Q-Mix software was developed by (28SantosRA, MeiraGR, BezerraWVDC, BragaFAV, PontesDL. 2021. Use of numerical method for optimization of granulometric curves in eco-efficient concrete. Matéria, Rio de Janeiro. 26(4):e13115. 10.1590/S1517-707620210004.1315.). The software contributes to the calculation of the optimal proportion of each aggregate fraction, optimizes the mixture of aggregates, and improves understanding the modified Andreasen packing model.

For a better comprehension, a step-by-step instruction for using Q-Mix to optimize packaging is presented.

Step 1: Feeding data obtained in the aggregate granulometry tests.

The user is asked to enter the data obtained through the granulometry test and name the respective aggregates.

Step 2: Selection of the input parameters.

To calculate the ideal granulometric curves using packing models, the software adopts modified A&A model; the user can select the minimum diameter (D_S), the maximum diameter (D_L), and distribution module (q) to be used.

The input data were the distribution modulus “q”, the largest and the smallest aggregate diameter (D_L and D_S, respectively). D_L and D_S values were fixed at 9.5 mm (gravel 9.5mm) and 0.075 mm (fine sand), respectively, which were obtained from the granulometric analysis of the aggregates in Q-Mix.

The optimization algorithm was developed through the linear programming, a mathematical modeling technique in which a linear function is maximized or minimized when subjected to various constraints. Using the individual granulometric curves (PSD) of each aggregate, the simulations (Q-Mix) calculated the optimal aggregate percentages required to plot a numerical PSD curve (mixture) with minimum deviation to its respective mathematical PSD curve (Modified A&A model).

In Table 4 it is presented the mathematical solution to the case of a mixture with “m” aggregates and “n” sieves/particle size classes.

Table 4 Calculation process of the deviation (RSS) between the numerical and the mathematical PSD curves for a mixture with m aggregates. 

$\mathrm{\Phi }$	$\mathrm{p}\mathrm{e}\mathrm{}\left(\mathrm{\%}\right)$	$\mathrm{p}{\mathrm{a}}_1\left(\mathrm{\%}\right)$	$\mathrm{p}{\mathrm{a}}_2\left(\mathrm{\%}\right)$	…	$\mathrm{p}{\mathrm{a}}_{\mathrm{m}}\left(\mathrm{\%}\right)$	$\mathrm{p}\mathrm{c}\left(\mathrm{\%}\right)$	$\mathrm{\Delta }$
${\mathrm{\Phi }}_1$	$\mathrm{p}{\mathrm{e}}_1$	$\mathrm{p}{\mathrm{a}}_{\mathrm{1,1}}$	$\mathrm{p}{\mathrm{a}}_{\mathrm{2,1}}$	…	$\mathrm{p}{\mathrm{a}}_{\mathrm{m},1}$	$\mathrm{p}{\mathrm{c}}_1$	${∆}_1=\mathrm{}\mathrm{p}{\mathrm{c}}_1-\mathrm{p}{\mathrm{e}}_1$
${\mathrm{\Phi }}_2$	$\mathrm{p}{\mathrm{e}}_2$	$\mathrm{p}{\mathrm{a}}_{\mathrm{1,2}}$	$\mathrm{p}{\mathrm{a}}_{\mathrm{2,2}}$	…	$\mathrm{p}{\mathrm{a}}_{\mathrm{m},2}$	$\mathrm{p}{\mathrm{c}}_2$	${∆}_2=\mathrm{}\mathrm{p}{\mathrm{c}}_2-\mathrm{p}{\mathrm{e}}_2$
…	…	…	…	…	…	…	…
${\mathrm{\Phi }}_{\mathrm{n}}$	$\mathrm{p}{\mathrm{e}}_{\mathrm{n}}$	$\mathrm{p}{\mathrm{a}}_{1,\mathrm{n}}$	$\mathrm{p}{\mathrm{a}}_{2,\mathrm{n}}$	…	$\mathrm{p}{\mathrm{a}}_{\mathrm{m},\mathrm{n}}$	$\mathrm{p}{\mathrm{c}}_{\mathrm{n}}$	${∆}_{\mathrm{n}}=\mathrm{}\mathrm{p}{\mathrm{c}}_{\mathrm{n}}-\mathrm{p}{\mathrm{e}}_{\mathrm{n}}$
$\text{\hspace{1em}RSS}$							$\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\sum {\mathrm{\Delta }}_{\mathrm{i}}^2$

Where:

${\Phi }_i$ – Sieve/particle size class “i”;

${\text{pe}}_i$ – Accumulated retained percentage in sieve “i”, calculated by the modified A&A model;

${\text{pa}}_{j,i}$ – Accumulated retained percentage in sieve “i” for aggregate$a_j$;

${\text{pc}}_i$ – Accumulated retained percentage in sieve “i” for the combination of aggregates 1, 2, …, m;

${∆}_i$ – Deviation between the accumulated percentage of aggregates retained in sieve “i” and its respective recommended accumulated retained percentage by the mathematical model;

RSS – Deviation between the numerical PSD and the mathematical PSD, calculated as a residual sum of squares, where: i varies from 1 to n, and j varies from 1 to m.

The packing optimization is performed through linear programming, in which the objective function to be minimized corresponds to the RSS (1), and the imposed restriction is $\sum _1^m{\beta }_j=\text{100\%}$. Therefore, the optimization consists in varying the coefficients βj until a minimum RSS value is found.

$\text{RSS}=\sum _1^n{\Delta }_i^2=\frac{(p{c}_i-p{e}_i{)}^2}{100}$ [1]  

Where:

${\text{pe}}_i=\frac{D_i^q-D_s^q}{D_L^q-D_s^q}\times \text{100}$ (Modified A&A model)

${\text{pc}}_i=\sum _j^m{\beta }_j{\text{pa}}_{j,i}$

Step 3: Run the software the mixture and obtain the optimized percentages.

After selecting the input parameters and running the algorithm, the modified A&A model will appear in the lower left corner of the screen. At this moment, the “solve” command must be clicked on.

To obtain more satisfactory results, the chosen aggregates must be of good quality, as the calculation of the software is based on the input data.

Figure 3 shows the Q-mix software screen, in which a random example is presented with data of three aggregates from the local commerce (Cajazeiras-PB/ Brazil).

  

Figure 3 Example of packing calculation for three chemical a content of aggregates in Q-Mix. 

2.5. Preparation, curing, and casting of the test specimens

Initially, all the materials necessary for preparing concrete were separated and temporarily stored. The materials included the Portland cement; limestone filler; fine, medium and coarse sand; small gravel; superplasticizer and water. The dry materials were stored in translucent plastic bags, and the superplasticizer and water were stored in beakers and buckets.

Table 5 summarizes the information on the consumption of materials used in concrete mixtures. The informed data correspond to an aggregate optimization process to determine the ideal percentage to ensure the best performance of these materials. The water/fines ratio was maintained at 0.44 for concretes with packaging, except for T160 concrete, which, due to the higher content of LF, required a greater amount of water for the desired slump of about 120mm.

The aggregate contents were calculated using Q-mix software (28SantosRA, MeiraGR, BezerraWVDC, BragaFAV, PontesDL. 2021. Use of numerical method for optimization of granulometric curves in eco-efficient concrete. Matéria, Rio de Janeiro. 26(4):e13115. 10.1590/S1517-707620210004.1315.), which allows us to determine the ideal proportion of materials considering the previously informed data.

Regarding admixtures, 1.8% of the total mass (cement) was used in the concretes with filler, except for concrete T160, which required 2.2% of superplasticizer. For reference concrete (T-RCE and T-RSE) 0.8% of superplasticizer was used.

The concrete preparation procedure started with the introduction of gravel and some amount of water in a 150 l concrete mixer. A few seconds after the equipment turned on, a half of the binder was added, followed by the intercalated pouring of all of the coarse sand to promote better mixing with the binder, especially in filler mixes. Then, most of the water was added to the mixer, leaving only a small amount in the bucket to mix with the chemical admixture if necessary. Then, medium and fine sand were added gradually. A few seconds after the mixture started rotating in the mixer, the rest of the binder was added. Finally, the additive was added to the concrete along with the remained water calculated and stored in the beakers.

The concrete was cast into cylindrical metal molds (dimensions: 10 × 20 cm). For each concrete mixture, 12 specimens were cast, which were mechanically compacted. After 24 h of cast, the specimens were demolded, identified, and submerged in a saturated lime water solution (25 ± 2°C) for 7 and 28 days to perform compressive strength tests.

2.6. Destructive and nondestructive tests

2.6.2. Chloride migration test

 ⌅

Tese tests were conducted according to the guidelines established in the procedure for specimen preparation and the migration test described in (35NT BUILD 492. 1999. Chloride migration coefficient from non-steady-state migration experiments.). This standard determines the migration coefficient of chlorides in concrete, mortar, and other cement-based material using a non-steady-state migration method. This coefficient is used as an indicator of the resistance of the material to chloride penetration.

After a period of curing while submerged in a tank containing saturated lime solution, two cylindrical samples (dimensions: 10 × 20 cm), aged 28 days, for each concrete mixture (T160, T240, T320, T-RSE and T-RCE) were removed from the tank. The specimens were divided into five parts: they were cut perpendicular to their longitudinal axis, and the extremities were discarded, while the remaining three parts were used in the migration test. Thus, for each specimen, three samples were selected, and thus, six samples were tested for each concrete mixture.

Before starting the test, the specimens were waterproofed in the entire lateral surfaces and saturated in lime water solution under vacuum. Then, the specimens were packed in PVC tubes to ensure that only the circular surface faces were in contact with the solutions.

Two different solutions, one cathodic and the other anodic, were prepared. The cathodic solution was stored in an acrylic container and contained 100 g of NaCl for every 900 g of water. This solution was connected to the negative pole of the power supply using a stainless steel electrode. The anodic solution contained 12 g of NaOH for each liter of water and was placed in a plastic tube above the samples. This solution was connected to the positive pole of the power supply also using a stainless steel electrode. Both solutions were maintained within a specific temperature range, usually 20°C–25°C. Figure 4 shows all the details of the NT Build test (35NT BUILD 492. 1999. Chloride migration coefficient from non-steady-state migration experiments.).

During the test, the specimens were subjected to a potential difference between the cathodic and anodic solutions. The test results were used to evaluate the resistance of concrete to the corrosion caused by chloride migration.

  

Figure 4 Assembly of the migration test set. 

After the testing period, specimens were splitted into two parts with a thickness of 50 mm. Then, a colorimetric indicator solution was sprayed. In this case, the concrete surface was brought into contact with a AgNO₃ solution, thereby initiating a chemical reaction that resulted in the formation of a white precipitate containing free chloride ions. Thus, the penetration of chlorides into concrete was measured. Afterwards, the migration coefficient was calculated acording to NT Build test (35) recomendations.

2.6.4. Wetting and drying cycles in NaCl solution

 ⌅

The test consists of subjecting the specimens to wetting and drying cycles in aggressive solution. The wetting semi-cycle was characterized by a total immersion in a 1M sodium chloride solution for 2 days and the drying semi-cycles was performed in an oven at 40ºC for 5 days (Figure 5). The adopted solution had a concentration commonly reported in the literature (38PageMM, PageCL, NgalaT, AnsticeDJ. 2002. Ion chromatographic analysis of corrosioan inhibitors in concrete. Constr. Build. Mater.16(2):73-81. 10.1016/S0950-0618(02)00017-X.-40SilvaFG. 2006. Estudo de concretos de alto desempenho frente à ação de cloretos. Tese de Doutorado em Ciência e Engenharia dos Materiais, Escola Politécnica, Universidade de São Paulo, São Carlos – SP.) and the characteristics of the drying period was similar to some previous works (39VieiraFMP. 2003. Contribuição ao estudo de corrosão de armaduras em concretos aramados com adição de sílica ativa. Tese de doutorado em Engenharia, UFRGS, Porto Alegre – RS., 41KishimotoI. 2010. Experimental study on the corrosion condition of steel bars in cracked reinforced concrete specimen. In: International Symposium on the Ageing Management & Maintenance of Nuclear Power Plants, 2010, Tokyo. Proceeding, Tokyo. 166-172.).

  

Figure 5 Wetting and drying test setup. 

2.7. Statistical analysis

The data experimentally obtained were analyzed using the statistical tool analysis of variance (ANOVA). This tool allows one to compare multiple groups, based on the F test, to establish the acceptance or rejection of the null hypothesis (Ho), which is determined through equality between the compared means. In addition, the Tukey test was performed for the data characterization process to identify differences and homogeneity of variance between groups. The significance level adopted for the study was 5%.

3.1. Compression Strength

Figure 6 presents the results of the axial compressive strength tests for each mixture. It can be seen the effect of an increasing percentage of LF mass on the compressive strength at 7, 28, and 90 days.

  

Figure 6 Compressive strength of the studied concretes at 7, 28, and 90 days. 

As expected, it was verified that the compressive strength increased with time and decreased with the addition of LF, except for the T240 concrete, which presents an intermediate LF content. When comparing the other mixtures containing filler with the T240 concrete, which exhibited the highest compressive strength at all ages and had a better performance than the reference mixture without packing at the ages of 28 and 90 days. It is important to emphasize that the mixtures T160, T240, T320, TRCE, and T-RSE reached the minimum compressive strength of 20 MPa at 28 days, as required by the Brazilian standards.

(49BederinaM, MakhloufiZ, BouzianiT. 2011. Effect of limestone fillers the physic-mechanical properties of limestone concrete. Phys. Procedia.21:28-34. 10.1016/j.phpro.2011.10.005.) concluded that, in some replacement ranges, the addition of LF caused an increase in the compressive strength of the concrete, whereas in other proportions, there was a decrease in this property.

Other studies also corroborate this trend of a reduction in compressive strength with an increase in the LF content in concrete. For example, (9CândidoTG, MeiraGR, QuattroneM, JohnM. 2022. Mechanical performance and chloride penetration resistance of concretes with low cement contents. Rev. IBRACON Estrut. Mater.15(6):e15608. 10.1590/S1983-41952022000600008.) reported a decrease in compressive strength with an increase in LF content in concrete. The authors concluded that increasing the LF content reduced the compressive strength because LF was less reactive than Portland cement.

The influence of LF content on compressive strength of concretes is more significant at higher replacement levels because, at lower levels, the packing of the cementitious matrix promoted by the filler compensates the reduction of cement hydration products (50WangD, ShiC, FarzadniaN, ShiZ, JiaH, OuZ. 2018. A review on use of limestone powder in cement-based materials: Mechanism, hydration and microstructures. Constr. Build. Mater.181:659-672. 10.1016/j.conbuildmat.2018.06.075.). Replacing part of the cement by LF increases the effective water to binder products and provides more space for hydration products, increasing the hydration degree of the binders. In addition, the filling effect and the heterogeneous nucleation of LF also contribute to the compressive strength increase. As a result, when small proportions of Portland cement are replaced by LF, there may be an increase in compressive strength (23SunJ, ChenZ. 2018. Influences of limestone powder on the resistance of concretes to the chloride ion penetration and sulfate attack. Powder Technol. 338:725-733. 10.1016/j.powtec.2018.07.041.). Figure 7 presents the results of the statistical analysis of the compressive strength results.

  

Figure 7 Tukey test for compressive strength of concrete. 

The results of the Tukey test showed a statistically significant difference in the compressive strength of the T-160 mix compared with all the other analyzed mixes (T-240, T-320, T-RCE, and T-RSE), indicating that T-160 has a different and lower compressive strength. Another important analysis was the comparison of the T-240 concrete with the other concretes; T-240 did not show any statistically significant difference to the T320 concrete, even though the latter contained 29.12% more Portland cement. This result indicates that this filler replacement proportion achieved the best matrix packing.

3.2. Binders intensity

Obtaining concrete with low environmental impact comprises much more than only replacing clinker by mineral additions. To produce more environmentally efficient concretes, it is necessary to reduce the binder consumption while increasing or maintaining the mechanical performance. Based on this premise, (51DamineliBL, KemeidFM, AguiarS, JohnM. 2010. Measuring the ecoefficiency of cement use. Cem. Concr. Compos.32(8):555–562. 10.1016/j.cemconcomp.2010.07.009.) proposed an indicator called binder intensity (Bi),; measured in units of kilogram per cubic meter), which evaluates the efficiency of concrete considering the amount of binder needed to reach a mechanical strength of 1 MPa. Thus, the Bi allows a more comprehensive analysis of the efficiency of concrete, considering both the environmental aspect and the mechanical performance.

where,

Bi = binder intensity, in kgm⁻³ MPa⁻¹;

C = cement consumption (kgm⁻³);

R = compressive strength (MPa).

For each concrete mixture analyzed, the Binder Index (Bi) was calculated considering the compressive strength at 7, 28, and 90 days, along with the respective cement consumption. Following the classification proposed by (52DamineliBL. 2013. Concepts for the formulation of low binder consumption concrete: rheological control, packing, and particle dispersion. São Paulo: Polytechnic School of the University of São Paulo.), which considers high-efficiency concretes as those with an Bi of up to 5 kgm⁻³ MPa⁻¹, at 28 days, the obtained values were compared to determine the efficiency of the studied concretes. Table 7 presents all the Bi results for all concrete mixtures.

From the data in Table 7, it can be seen that the Bi decreases as the LF content increases. In the particular case of T240 concrete, Bi assumed the lowest values, showing that an optimal percentage of cement replacement by filler can be reached to concrete achieve a good binder index. This result shows that T240 concrete presented a better performance than the reference concrete using packing procedures.

Figure 8 shows the Bi results extracted from the literature and the concrete results of this study (9CândidoTG, MeiraGR, QuattroneM, JohnM. 2022. Mechanical performance and chloride penetration resistance of concretes with low cement contents. Rev. IBRACON Estrut. Mater.15(6):e15608. 10.1590/S1983-41952022000600008., 31GraziaMT, SanchezLFM, RomanoRCO, PileggiRG. 2019. Investigation of the use of continuous particle packing models (PPMs). On the fresh and hardened properties of low-cement concrete (LCC). systems. Construct. Build. Mater.195: 524–536. 10.1016/j.conbuildmat.2018.11.051., 53SuN, HsuKC, ChaiHW. 2001. A simple mix design method for self-compacting concrete. Cem. Concr. Res.31(12):1799–807. 10.1016/S0008-8846(01)00566-X.-63LopesHMT, PeçanhaACC, CastroAL. 2020. Considerações sobre a eficiência de misturas de concreto de cimento Portland com base no conceito de empacotamento de partículas. Revista Matéria. 25(1). 10.1590/S1517-707620200001.0874.). It can be observed that the studied concretes present good results compared to other works that studied different concretes with different types of mineral additions.

  

Figure 8 Relationship between binder intensity and compressive strength at 28-days - literature and experimental data. 

3.3. Chloride migration

Non-steady-state accelerated migration tests were performed at 28 and 90 days of age for the studied concretes, following the procedures described by (35NT BUILD 492. 1999. Chloride migration coefficient from non-steady-state migration experiments.). The migration coefficient values obtained for each experimental condition at their respective ages are shown in Figure 9.

  

Figure 9 Migration coefficients for the studied concretes. 

The results showed that the partial replacement of Portland cement by LF increased the chloride migration coefficients ions at all ages compared to the reference mixtures.

Furthermore, there was a significant reduction in the migration coefficients in concretes with advanced curing ages. For example, concrete T160 was 25.92% more resistant to chloride ingress at 90 days than at 28 days of curing. In the mixtures with higher cement contents, chloride migration coefficient decreased, but the reduction was less significative than that observed in the first test at 28 days.

The results obtained in this work are in agreement with the results reported by (23SunJ, ChenZ. 2018. Influences of limestone powder on the resistance of concretes to the chloride ion penetration and sulfate attack. Powder Technol. 338:725-733. 10.1016/j.powtec.2018.07.041.). These authors argued that chloride penetration into concrete were deeper in concretes containing LF than in concretes that did not have this kind of mineral addition. These authors explain that the deeper penetration of chloride ions in these concretes are related to the higher level of OH⁻ ions present in the pore fluid of the concrete made with LF, and the characteristics of the porous and connected paste–aggregate interfacial transition zone (ITZ) associated with LF particles addition. (64Andrade, C. 1993. Calculation of chloride diffusion coefficients in concrete from ionic migration measurements. Cem. Concr. Res.23(3):724-742. 10.1016/0008-8846(93)90023-3.) suggested that the OH⁻ ions present in the pore fluid act as a supporting electrolyte and are responsible for the transportation of a significant amount of charge during chloride ion permeability test because of its higher ionic conductivity than the other ions present in pore fluid (Na⁺, K⁺ and Ca²⁺). (65LunaFJ, FernándezÁ, AlonsoMC. 2018. The influence of curing and aging on chloride transport through ternary blended cement concrete. Mater. Construcc.68(332):e171. 10.3989/mc.2018.11917.) reported a reduction in chloride migration coefficients when a blast furnace and LF were used However, they used much less LF than that used in this research.

Considering the classification presented in Table 8, T-RCE and T-RSE concretes demonstrated moderate resistance to chloride migration, whereas the T-320, T-240, and T-160 concretes showed low resistance at 28 days.

Chloride migration coefficients were correlated with other concrete study parameters, such as compressive strength and cement consumption. The results obtained are presented in Figures 10 and 11.

  

Figure 10 Correlation between chloride migration coefficients and compressive strength. 

  

Figure 11 Correlation between chloride migration coefficients and cement content. 

When comparing the results of chloride migration coefficients with the compressive strength and cement content, it was observed that cement content correlates better with migration coefficients. However, chloride migration coefficient also presents good correlation with compressive strength (R² = 0.8064).

(67MehtaK, MonteiroJM. 2014. Concrete: microstructure, properties, and materials. 4th ed. New York: McGraw-Hill Education.) stated that compressive strength is directly related to concrete durability, including the reduction of the chloride migration coefficient. They explained that a denser and more resistant concrete matrix is less porous, which makes it difficult for chloride ions to penetrate.

By analogy, the graph in Figure 11, which correlates cement content to the chloride migration coefficient, also reinforces the conclusion that the greater the cement content, the less easy it is for chlorides to penetrate into the concrete, which presents relation with the increasing chloride binding ability of concretes with higher cement content.

3.4. Capillary absorption

Water capillary absorption by the capillarity of cylindrical specimens (C) is calculated using the following Equation:

Where Msat (g) is the saturated mass of specimen, Ms (g) is the mass of the dry specimen and S is the cross-section area (cm²).

When plotting a graph with the values of C (g.cm⁻²) against the square root of time, a linear relationship is observed, as shown in Figure 12. The slope of this curve is the coefficient of capillary absorption. The mixtures containing an adequate proportion of LF resulted in good absorption results. However, the T160 concrete, which contains a high amount of filler, exhibited greater water absorption coefficient than the T320 concrete, which was about 40% higher. Furthermore, particle packing, which determines the filling of voids, greatly affected the result of capillary absorption. The T-RCE concrete presented the best results in the water capillary absorption test, which presented a capillary absorption raten 56% lower than the reference mixture without packing.

  

Figure 12 Correlation between capillary absorption and square root of time. 

The correlation between cement consumption and capillary absorption coefficient is better represented in Figure 13. Concrete with a high cement replacement rate, and consequently, lower content of this binder, yielded high absorption values. The Figure 13 also highlights the difference between concrete with packaging (T-RCE)and conventional reference concrete (T-RSE).

  

Figure 13 Correlation between capillary absorption coefficient and cement content. 

3.5. Electrochemical monitoring

Figure 14 shows the results of electrochemical monitoring of the steel bars embedded in concrete.

  

Figure 14 Electrochemical monitoring of the steel bars embedded in concrete (average results). 

As expected, the steel bars embedded in T160 concrete, which had the lowest cement consumption, were the first to depassivate, requiring 119 days under wetting and drying cycles. T160 was closely followed by the reference concrete without aggregate packing (T-RSE) that showed a shift in the current density measurements after 133 days. Even with 57.19% less cement, T160 concrete presented a performance close to that of the conventional concrete (T-RSE). Figure 15 shows the reinforcement depassivation times for all concretes.

  

Figure 15 Depassivation time of bars embedded in concrete. 

(44MeiraGR, FerreiraRR, JerônimoL, CarneiroAMP. 2014. Comportamento de concreto armado com adição de resíduos de tijolo cerâmico moído frente à corrosão por cloretos. Ambiente Construído. 14(4):33–52. 10.1590/S1678-86212014000400004.) studied concretes with ceramic residues (CR) and electrochemically monitored bars embedded in concrete with 10% and 30% of cement replacement by this residue, resulting in cement consumptions of 372.9 kg.m⁻³ and 290 kg.m⁻³, respectively, and a reference concrete with cement consumption of 414.3 kg.m⁻³. The results of the electrochemical monitoring show that the passivation of the reference bars occurred in approximately 40 days, while that of the concrete with 10% RTM occurred in approximately 28 days and that of the concrete with 30% RTM occurred in 42 days. (68JerônimoL, MeiraGR, FilhoLCS. 2018. Performance of self-compacting concrete with wastes from heavy ceramic industry against corrosion by chlorides. Constr. Build. Mater.169:900-910. 10.1016/j.conbuildmat.2018.03.034.) monitored self-compacting concrete with an average cement consumption of 515 kg.m⁻³. Despite the high cement consumption, the bars depassivated after 50 days. Therefore, we can infer that the packaging of the aggregates applied in concretes T160, T240, and T320 compensated for the reduction in the cement content, thereby retarding the activation of rebars in these concretes.

The correlations shown in Figure 16 explain the effect of packaging on the initiation time of corrosion by chlorides. In the electrochemical monitoring, for the group of concretes with identical granular skeletons, cement content and chloride migration coefficient were strongly correlated (Figure 16 (a)). Since this also depends on the cement as the greater the clinker content, the greater is the aluminate content that can combine with the chlorides and fix them in the matrix, and in turn, longer is the depassivation time (69RibeiroDV, LabrinchaJA, MorelliMR. 2012. Effect of the addition of red mud on the corrosion parameters of reinforced concrete. Cem. Concr. Res.42(1):124–133. 10.1016/j.cemconres.2011.09.002.-71TianL, DaiS, YaoX, ZhuH, WuQ, LiuZ, ChengS. 2022. Effect of nucleation seeding and triisopropanolamine on the compressive strength, chloride binding capacity and microstructure of cement paste. J. Build. Eng.52:104382. 10.1016/j.jobe.2022.104382.). Figure 16 (b) shows that there was no considerable correlation when comparing the concrete with packing with the reference concrete, even when the latter had a higher cement content. Thus, it was confirmed that the packing has a significant role in corrosion initiation retardation.

  

Figure 16 Correlations between cement content, depassivation time, and chloride migration coefficient, considering only packed concretes (a) and (b) all concretes.