Effect of early-age freeze-thaw exposure on the mechanical performance of self-compacting repair mortars

2.1 Materials and mix proportions

The physical, chemical, and mechanical characteristics of Portland cement (CEM I 42.5 R) used in this study are presented in Table 1. Crushed limestone sand with a maximum size of 4 mm was used as fine aggregates. The specific gravity and water absorption properties of crushed sand were 2.62 and 1% (by mass), respectively. The chemical composition of mineral powders is listed in Table 2. According to ASTM C 618, FA can be classified as Class C. FA was procured from the Soma power plant, which is one of the biggest power plants in Turkey. GGBFS can be classified as Grade 100 in accord with ASTM C989. The specific gravity and Blaine fineness of FA were 2.06 and 290 m²/kg, respectively. The specific gravity of GGBFS was 2.9, and the Blaine fineness of GGBFS was 396 m²/kg. LP was used as a viscosity-enhancing material. LP was a byproduct of crushed limestone aggregate with a specific gravity of 2.65 and a Blaine fineness of 440 m²/kg. A new-generation polycarboxylate-based SP meeting standard specifications of ASTMC 494 Type F was used in this study. The solid content, pH, and specific gravity of the admixture were 40%, 6.5, and 1.10, respectively. A straight-type steel fiber (Dramix OL 6/16) was used with 78.5 kg/m³ (1 vol.%) fiber dosage. The microfiber was made of high-strength steel with a brass coating, which is the most suitable for SCRM. Its tensile strength, length, and diameter were 2250 MPa, 6 mm, and 0.16 mm, respectively.

The primary aim of mix proportion was to achieve 60±5 MPa compressive strength at 70 days with or without replacement. This strength level was chosen to reach the desirable durability performance and bond properties between the repair mortar and the existing concrete. Therefore, 14-day strengths (initial strengths for F-T exposure) were different for all mixtures for the purpose of evaluating early-age effect. Cement was replaced with FA and GGBFS at 40% and 60% (by mass), respectively. To provide a constant water/binder ratio and to reach target spread value (300 mm for nonfiber mortars) without segregation, SP was used in different dosages for each binder type. Mix proportions are presented in Table 3. Note that the following total cementitious content was kept constant at 500 kg/m³ and the water/powder ratio at 0.29. LP was used at a constant amount for all mixtures (200 kg/m³). An increase in paste volume due to GGBFS or FA was balanced with decreasing aggregate content. Note that the abbreviations were used in tables and figures: C (control mixture without GGBFS or FA), C-SF (control mixture with steel fiber), FA (40% FA replacement), FA-SF (FA mixture with steel fiber), GGBFS (60% GGBFS replacement), and GGBFS-SF (GGBFS mixture with steel fiber).

SCRM mixtures were prepared in a vertical axis mixer. A more complicated mixing operation compared to the conventional mortar was applied. First, fibers and aggregates were mixed and powders (cement, GGBFS, FA, and LP) were added to the mix. After remixing, water was added to the dry mix. Finally, SP was introduced to the wet mixture. In the fresh state, slump-flow diameter and V-funnel time of the SCRM mixes were measured with a mini-apparatus in conformity with EFNARC [10] standards.

2.2 Test methods

After 24 h curing in a cabinet with 20±1°C and 100% RH, all prismatic specimens (40×40×160 mm) were demolded and submerged in lime-saturated water at 20°C. After 7 days of water curing and 7 days of drying period at 23±2°C with 50% RH, 56 F-T cycles were applied to the specimens for 56 days. This moderate drying period allowed the absorption of salt solution while thawing. Oven drying was not applied due to its important damages for cement-based material microstructure in the capillary porosity domain [11]. The saturated specimens were frozen at -20°C for 12 h and thawed in 10% sodium chloride (NaCl) solution at 20°C for 12 h. NaCl solution was used to accelerate the effect of F-T and to demonstrate the salt scaling. Chloride penetration depths were also measured on prismatic specimens after 56 F-T cycles using 0.2% AgNO₃ solution sprayed on freshly broken pieces left from flexural tests.

Prisms (25×25×285 mm) were used to determine the shrinkage potentials of each SCRM mixture according to ASTM C596 [12]. Four specimens were measured and the average values were recorded. All length changes were measured by a 1/1000 mm mechanical comparator. Free drying shrinkage measurements were initiated after 24 h curing in a cabinet with 20±1°C and 100% relative humidity (RH). The climate conditions for all specimens featured a temperature of 23±2°C and an RH of 50% to investigate drying shrinkage for 70 days.

To evaluate F-T resistance, F-T-exposed mortars were compared to the same-age water-cured specimens at the end of the cycles. That is, comparisons were done between specimens having similar maturity. Four prismatic specimens were used to determine the flexural strength, fracture energy, and compressive strength of mixtures exposed and nonexposed to freezing and thawing.

The flexural and compressive strength tests were performed according to ASTM C348 [13] and ASTM C349 [14], respectively. All flexural strengths and load-deflection graphs were obtained with carrying out three-point bending tests by an electromechanic closed-loop testing system (loading rate, 0.02 mm/min). After cycles (at 70 days), the specimens were loaded from their mid-span, and the clear distance between simple supports was 130 mm (Figure 1). The beam specimens had the same notch depth for all series, equal to a quarter of the beam height. The fracture energy was determined using the load-deflection curves of the three-point bending test results. The fracture energy (G_F) was calculated by dividing the area under the load-deflection curve by the effective cross-section area for each specimen. The mid-span deflection was up to 2 mm for fiber-reinforced series. In this study, small prismatic samples were used in flexural loading. Therefore, the weight of the specimens was neglected in the calculation of fracture energy. Compressive strength tests were applied on two pieces left from flexural tests with a 2400 N/s loading rate.

Figure 1

Schematic presentation of three-point bending test.

3.1 Fresh state properties

The mini V-funnel and mini slump-flow test results are presented in Figure 2. The workability tests results showed that GGBFS usage decreased SP demand to reach the target spread diameter and V-funnel time in spite of its higher Blaine fineness compared to FA and cement. This can be attributed to the nonporous nature and smooth surface of GGBFS. FA, which has the lowest Blaine fineness, demonstrated worse performance about SP saving. Despite its spherical body and lower Blaine fineness, FA had more SP absorption capacity compared to others. A high carbon content in FA may adversely affect workability and lead to erratic behavior with respect to chemical additives becoming adsorbed by the porous carbon particles [15]. In Figure 2, the steel microfiber addition decreased the workability of control and FA mixtures, whereas GGBFS mixtures were not affected considerably. The steel microfibers created an additional inner friction and exhibited less workability in the same SP dosage compared to plain mortars. However, all mixtures showed great filling performance, including fiber-reinforced ones.

Figure 2

Mini V-funnel and slump-flow test results.

3.2 Compressive strength of SCRM

Figure 3 shows the compressive strength of SCRM mixtures at different ages and after different exposure conditions. The initial strength was determined after 7 days of water curing and 7 days of air curing at 20°C. The residual strength was determined after 56 F-T cycles (sample age, 70 days). The compressive strength test was also performed on water-cured samples (20°C) at the same age (70 days). In Figure 3, high-volume FA or GGBFS replacement reduced the compressive strength of SCRM at early ages (14 days). This finding was also valid for SCRM containing steel fibers. The compressive strength reduction was 17% for 40% FA replacement compared to the Portland cement SCRM (control) at 14 days. This decrement was 35% in the case of steel microfiber-reinforced SCRM containing FA. The compressive strength of SCRM containing GGBFS was 18% lower than the control mixture at 14 days. This decrement was also 18% for fiber-reinforced SCRM. That is, high-volume FA or GGBFS replacement reduced the early-age strength as expected. The strength reduction in GGBFS replacement was lower than in FA replacement. The normalized strength (FA or GGBFS SCRM/Portland cement SCRM) increased after 70 days of water curing (Figure 3). In this case, there was no strength reduction in using FA. Some decrement (16%) was also observed for FA mixtures containing steel fibers. High-volume GGBFS replacement did not reduce the compressive strength (8–10% decrement) significantly at later ages (70 days). From the point of compressive strength, prolonged curing is necessary in high-volume mineral additive replacement.

Figure 3

Initial, in water, and residual compressive strengths.

To evaluate the effect of F-T on SCRM, normalized strengths were used (Figure 4). In Figure 4, the compressive strength after 56 F-T cycles was generally over the initial strength (14 days) for all mixtures. This can be explained by the curing effect of 20°C NaCl solution during the thawing process. On the contrary, the compressive strength after F-T cycles was generally lower than the compressive strength of samples kept in water (at the same age), especially in the case of replacement. There was no compressive strength reduction for the control mixture (without mineral additives) after F-T cycles. In Figure 4, the compressive strength reduction due to F-T exposure was 5% for the control mixture containing steel microfibers, whereas the compressive strength reduction was 28% for FA mixtures. When cement was replaced with 60% GGBFS, the compressive strength loss due to F-T exposure was between 14% and 18%. These results may be explained by the curing age at the beginning of exposure time, which may be sufficient for the control mixture but may not be sufficient for high-volume mineral additive containing mixtures. It seems that steel microfibers did not change the F-T resistance behavior from the point of compressive strength. This can be attributed to the weak transition zone between fibers and matrix at early ages. That is, F-T exposure at early ages may cause bond damage between the fiber and the matrix.

Figure 4

Normalized compressive strength after F-T exposure compared to the initial and water-cured strength.

A weaker resistance to deicer salt scaling for GGBFS as cement replacement compared to ordinary Portland cement concrete when tested at 28 days was also reported [16, 17]. The reasons for the poorer performance are not clear. However, some researchers have partially attributed it to the slow hydration and insufficient curing regime for concrete containing GGBFS [16, 17]. Gutiérrez et al. [18] researched the influence of pozzolans on the performance of fiber-reinforced mortars and reported that, in a general manner, pozzolanic materials, particularly silica fume and metakaolin, improve the mechanical performance and durability of fiber-reinforced materials and that FA addition had a poor performance, which was associated with its low degree of pozzolanic activity. Pigeon et al. [19] investigated the effects of microfibers on frost damage. The results of the freezing and thawing cycle tests and of the deicer salt scaling tests showed that the microfibers reduced the rate of deterioration due to freezing phenomena but did not completely prevent damage.

3.3 Flexural performance of SCRM

The load-deflection relationship of mixtures containing steel microfibers exposed and nonexposed to frost attack, including NaCl, is presented in Figure 5. In Figures 5 and 6, steel microfibers changed their behavior completely compared to the mixtures having no fiber under flexural loading. The gradual load decrement was observed in all series after the peak load. High postpeak load-carrying capacity shows well toughness and reinforcing effect of the steel fibers. The formation of strain hardening after the first crack is a typical indication of high performance. In Figure 5, all SCRM mixtures containing steel fiber showed strain hardening behavior. The residual strength at 2.0 mm deflection was also a significant level (for control and FA mixtures, 65–70% of maximum load; for GGBFS mixture, 50% of maximum load). In Figure 5, F-T exposure reduced the peak load; however, the general character of the curves was not affected. The maximum load was obtained from the control mixture following the FA and GGBFS mixtures in water curing. This trend was the same after the effect of F-T, but the difference between the curves increased. This indicates that high-volume mineral additives increased the F-T sensitivity at the age of 14 days; this behavior did not change using steel fibers significantly. That is, early-age F-T exposure affected the fiber-matrix bond strength negatively.

Figure 5

Flexural load-deflection curves of fibered SCRM series.

Figure 6

Flexural load-deflection curves of plain SCRM series that were water cured and exposed to F-T.

The load-deflection relationship of non-fiber-reinforced mixtures is presented in Figure 6. As seen from the curves, plain mortars exhibited brittle behavior under flexural load, as expected. The graph area under final deflection is clearly lower than fiber-reinforced mortars. The general behavior did not change considerably after F-T cycles, but descending parts of the graphs shifted slightly to the right compared to the graphs of water-cured specimens. This can be attributed to the microcracks due to F-T exposure.

Figure 7 shows the flexural strength of SCRM mixtures at different ages and after different exposure conditions. Steel microfibers markedly increased the flexural strength for all mixtures. Fiber reinforcement was induced to increase the flexural strength within the range of 54%–72% and 65%–100% at the initial and 70-day curing periods, respectively. Fiber reinforcement provided the greatest flexural strength gain with prolonged curing time (14–70 days) in FA-replaced matrix. These increments induced by fiber reinforcement were found to be 54% for 14 days and 100% for 70 days. Hence, FA has lower pozzolanic activity compared to GGBFS, and portlandite might be replaced by calcium silicate hydrates at the fiber-FA matrix interface in a long time. Furthermore, these improvements are more pronounced using steel fiber. This can be explained by the improved bond strength between the fiber and the matrix.

Figure 7

Initial, in water, and residual flexural strengths.

Figure 8 shows the effect of F-T on flexural strength. The aggressive environment reduced flexural strengths compared to both 14- and 70-day-old specimens. F-T exposure caused flexural strength decrement up to 22%, except C and C-SF mixtures. Fiber reinforcement caused higher strength losses compared to nonfiber case. The fiber-matrix bond was adversely affected by F-T exposure. In fact, several researchers found that steel microfibers enhanced the frost resistance of concrete due to the air-entraining and crack-limiting effects of fiber incorporation [19, 20]. These effects of fiber reinforcement might be seen after prolonged curing, which improved the fiber-matrix bond considerably.

Figure 8

Normalized flexural strength after F-T exposure compared to the initial and water-cured strength.

The fracture energy of the mixtures is shown in Figure 9. Replacing cement with high-volume mineral additives (40% with FA and 60% with GGBFS) reduced the values of fracture energy of fiber-reinforced mortars after 70 days of water curing. A similar trend also existed after F-T exposure.

Figure 9

Fracture energy values for water curing and F-T exposure.

The positive effect of steel microfiber reinforcement on fracture energy was adversely affected by F-T exposure. Using fibers in control mixture increased the fracture energy 75 times compared to the control matrix without fibers. These ratios were 91 and 50 times for GGBFS and FA, respectively (Figure 10). However, in Figure 10, these increases decreased due to F-T exposure (30% for control mixture, 42% for FA replacement, and 1% for GGBFS replacement), which means that the fiber-matrix bond was adversely affected.

Figure 10

Normalized fracture energy of fiber-reinforced SCRM compared to the plain matrices without fibers.

In Figure 11, the fracture energies of fiber-reinforced series were affected more than its plain condition. Moreover, it is surprising that the control and FA mixtures without fibers exhibited an increasing tendency in fracture energy as 26% and 43%, respectively. This may be due to the crack branching or energy absorption of these random microcracks in the transition zone. However, the general behavior of the load-deflection curves did not change considerably due to F-T exposure. The flexural strength loss of steel microfiber-reinforced mortars due to F-T cycles was greater than plain mortars, so a similar behavior was valid for the fracture energy results.

Figure 11

Normalized fracture energy after F-T exposure compared to the fracture energy of water-cured samples.

3.4 Shrinkage tendency of SCRM

Figure 12 shows the drying shrinkage behavior of the SCRM mixtures. In Figure 12, replacing cement in the matrix phase of SCRM by mineral additives in high volumes affected the shrinkage behavior. GGBFS replacement increased the shrinkage slightly, whereas FA replacement decreased the shrinkage significantly compared to the control mixture. This behavior can be attributed to the reducing cement content and free CaO content of FA, which may compensate the shrinkage by expansion. In addition, the increase of the fineness of the pozzolanic material may give an increase in drying shrinkage. Because the fineness of GGBFS is higher than FA and cement, the use of GGBFS increased the shrinkage values significantly. Pozzolanic material usage against shrinkage may cause a positive or negative influence, especially in drying conditions [21]. Steel microfibers also reduced the shrinkage in all series. When concrete shrinks, these fibers restrain the shrinkage strain and thus reduce it. Therefore, the more fiber contents are used in the concrete, the more reduction in shrinkage strain is expected [7, 22]. Due to shrinkage cracks (especially at early ages), corrosive materials can penetrate into the concrete through the cracks and may damage rebars and the concrete body, too [23]. For this reason, the reduction in shrinkage may provide more durable repaired elements.

Figure 12

Drying shrinkage tendency.

3.5 Chloride penetration

The chloride penetration depths were measured on prismatic specimens after freezing and thawing using 0.2% AgNO₃ solution sprayed on freshly broken pieces. The chloride-contaminated area revealed a gray color, whereas the chloride-free area had dark brown appearance. The gray colorimetric front indicated the existence of free and water-soluble chlorides. The average depths of three specimens for each mortar are presented in Figure 13.

Figure 13

Chloride penetration depths.

GGBFS usage hindered chloride salt penetration into the mortar body, whereas FA had a bad performance. Steel microfibers did not affect the general trend, but only FA-SF exhibited poor performance compared to its nonfiber condition. The reduction in chloride penetration depth may be explained by the chloride-binding capacity of GGBFS and increasing resistivity. GGBFS was found to significantly enhance the pore structure of cementitious composites, increase its chloride-binding capacity (by creating more Friedel’s salt), and reduce its chloride diffusivity [24, 25].

SCRM specimens did not exhibit remarkable surface scaling at the deduction of the F-T cycling. On the contrary, chloride-induced corrosion mostly occurred in the chloride-penetrated area of the FA-SF mixture. Depending on visual observations (Figure 14), FA replacement caused more prominent rust spots in deeper areas of the cross-section, whereas the control mixture did not reflect corrosion products. The GGBFS-SF mixture exhibited very limited amount of corrosion spots, which were close to the surface. Kosa and Naaman [26] studied the corrosion of steel fiber-reinforced concrete in 3% NaCl solution and found that, for a certain level of exposure or degree of corrosion, the degree in toughness was much more dramatic than the decrease in strength. It was also noted that the strength and toughness decreased with an increase in the degree of corrosion, and these mechanical properties were primarily affected by the reduction in minimum fiber diameter. In this study, steel microfibers with only 0.16 mm diameter were used. Thus, a little decrease in fiber diameter or fiber-cement matrix bond due to corrosion may cause reduction in strengths and fracture energies.

Figure 14

Cross-section (40×40 mm) of fibered SCRM series after F-T exposure.

Above all, repair mortars with high-volume FA may not provide a sufficient cover to protect rebars against chloride corrosion even if a 20 mm repairing cover is applied with 7 days of water curing. Prolonged curing may be an effective method to overcome this problem.