The deterioration of existing concrete structures has led to the development of new rehabilitation systems, traditionally based on steel-plate or reinforced-concrete retrofitting, which, however, have the drawbacks of being difficult to implement and costly in terms of handling and material placement (1).

Fibre-reinforced polymers (FRPs) have excellent strength/weight ratios and anti-corrosive properties and have been successfully used in the automobile, aeronautical and sports sectors (2, 3). They have also been usefully applied in the construction industry, as evidenced by a large number of publications (4–7) and the development of the corresponding international regulations, standards and documentation (3, 8–10).

In most FRP studies of concrete confinement, carbon, glass and aramid fibres have been used with medium-resistance concretes subject to centred compression (11–17) or eccentric loading (1, 18, 19). Confinement studies of high-resistance concretes have also been published (20–23). It has been found that confinement effect is greater when concrete strength is lower (24–27) and that cyclic testing reduces rigidity (28–33). However, confinement in low-resistance concretes has not been sufficiently studied.

International standards and documentation regulating the execution and design of FRP reinforcements specify that implementation must be done with care. In many real in situ interventions, however, meticulous implementation to the level of care that is possible in laboratories cannot be guaranteed. Lacking in the literature are scientific studies of the impact of poor implementation on the behaviour of FRP-reinforced elements.

When a concrete component reaches compression stress of approximately 75% of the fracture stress, the crack system becomes unstable and internal microcracks develop in the mortar matrix (34). For values close to the 75% stress level, the available internal energy exceeds the required crack-release energy, thereby increasing the propagation rate and destabilizing the system. Due to the strength increase in confined concrete compared to the original (unconfined) concrete, it may happen that the service stress of the confined elements exceeds 75% of the ultimate strength of the unconfined concrete, resulting in a noticeable decrease in the rigidity of the elements.

Our threefold aim was to explore the behaviour of low- and medium-strength concrete specimens experimentally tested to compression and confined with carbon FRP (CFRP), as follows: (1) compare the behaviour of the two types of confined concrete; (2) study the behaviour of confined concrete when the concrete is already cracked; and (3) examine the impact of defective implementation conditions on the behaviour of reinforced specimens.

Three types of concrete were analysed. C1 was a low-strength concrete (8.5 MPa), analysed as an example of concrete manufactured with a high water-cement ratio. C2 and C3 were two commonly used medium-strength concretes (20.4 MPa and 25.2 MPa, respectively). Table 1 shows the proportions of each concrete type. The cement used was CEM II/A-L type 42.5 R manufactured by Lafarge and the maximum size of the coarse aggregate was 12 mm.

Six cylindrical specimens, diameter 150 mm and height 300 mm, were prepared of each concrete type, manufactured according to EN 12390-2 (35). The 18 specimens were first cured in a humid chamber at 20°C and 95% humidity for 28 days and then maintained in laboratory conditions at 20°C and 65% humidity for seven days. Specimen surfaces were polished to ensure that the press acted on smooth and plane surfaces perpendicular to stress direction during compression testing. Two specimens of each type of concrete were not reinforced so that they could be used as benchmarks. The remaining specimens were strengthened in two distinct conditions (50% and 100% CFRP reinforcement).

Before reinforcement, adhesive plastic sheets, some 50 mm wide, were glued vertically onto half of the specimens to be strengthened. These sheets were evenly distributed in such a way that 50% of the outer surface of the specimen was coated with the adhesive sheet, thereby preventing further CFRP bonding with the concrete. A primer (Drizoro XPS-400) was applied to all the reinforced specimens to improve surface resistance conditions (by closing pores and surface cracks) and to increase bonding with the carbon-fibre fabric (Replark MM2 Drizoro), glued onto all the specimens after 24 hours using epoxy resin (Drizoro XL-800) (Figure 1).

A first coat of epoxy resin adhesive was applied to bond the fibre tissue to the specimens. A 10 cm overlap was left to allow for removal of the fibre tissue when subjected to stress. Air bubbles and imperfections were subsequently removed with a roller. After 15 minutes (that is, once the coated surface was dry), a finishing coat of the same resin was applied. Two kinds of reinforced specimens were prepared: perfectly reinforced specimens (100% of the outer surface of the test specimen bonded to the CFRP) and partially reinforced specimens (50% of the outer surface of the test specimen bonded to the CFRP). After reinforcement, to complete the process of the resin polymerization, the specimens were left for 15 days in laboratory conditions (temperature 20°C and humidity 65%). The mechanical properties of the composite reinforcements are summarized in Table 2.

All the specimens underwent three cycles of loading and unloading up to 40% of the estimated ultimate strength (Figure 2). A code of the format CXYZ was used to identify the specimens, as follows: CX indicates specimen type (C1, C2 or C3); Y indicates test conditions (W: without reinforcement; M: 50% fibre-reinforced surface; T: 100% fibre-reinforced surface); and Z — a number between 1 and 18 — identifies the individual test piece.

Specimen fracture strength without reinforcement was obtained from previous tests. The ultimate fracture for each reinforced specimen was estimated in accordance with the Lam and Teng model given by Equation [1] (36, 37) as adopted by the American Concrete Institute (ACI) (3). f c c = f c + 6.6 ψ f E r n t f e k e ε f u D

where

Regarding the coefficients, ψ_f=0.95 comes from ACI Committee 440 (3). k_e accounts for failure of the reinforcing material to achieve ultimate strain (38) due to multiaxial stress at the location of the reinforcement material and the stress concentration caused by the concrete cracking as it dilates. Values obtained experimentally for k_e range from 0.58 to 0.61 (36, 39, 40). The value used in this study was the ACI (3) value of 0.55.

Tests were conducted, controlling strain, at a speed of 10-5ε/s, using a universal Ibertest MIB press with a load capacity of 1200 kN. Vertical displacement was recorded by means of two linear variable differential transformer (LVDT) displacement transducers. Horizontal movement was controlled by means of an MTS 632.11F.20 extensometer. Results were represented as longitudinal and transversal stress-strain curves. Longitudinal elastic modulus and Poisson's ratio values were calculated. Specimens were subsequently subjected to compressive strength testing up to fracture, in accordance with EN 12390-2 (35), so as to obtain the ultimate strength values.

Table 3 shows results for the three loading cycles and compressive testing up to fracture.

The first six columns show elastic modulus (E) and Poisson's ratio (υ) values for each loading cycle. The values are those obtained for a stress equal to 40% of the estimated ultimate fracture. The seventh column shows values for ultimate failure stress (σ_r) and the last three columns show mean elastic modulus, Poisson's ratio and ultimate strength values for each set of specimens and the specimen conditions on testing. The elastic modulus and Poisson's ratio values were obtained from the third loading cycle data (41). It should be noted that the standard refers to concrete specimens, not to FRP-confined concrete specimens.

Specimens failed (Figure 3) when the CFRP fractured perpendicularly to the fibre. No noticeable visual differences were evident for the two reinforcement levels (50% and 100%) or the three specimen types (C1, C2 and C3). Fracturing was preceded by sounds that indicated initial fracturing of the fibres. Satisfactory concrete-CFRP bonding was in all cases indicated by the attachment of a thin layer of concrete to the fibre. Fracturing occurred in areas of overlap between fibres.

In the first loading cycle—with stress values higher than service stress values—the behaviour of confined and unconfined specimens was very similar, except for a slight increase in rigidity in the former. No significant differences in behaviour were observed in the second and the third loading cycles.

When an unstable concrete microcrack level was reached in the first loading cycle (as happened with C1), rigidity sharply decreased in subsequent cycles. When stress increased, rigidity also increased slightly; likewise, the elastic modulus value also increased slightly as stress increased.

The percentage strength increase due to confinement was far higher in the low-resistance concrete than in the medium-resistance concretes.

Reinforcement resulting from concrete confinement was unaffected by defective implementation. Even for fibre tissues bonded to only 50% of the outer surface, the maximum strength decrease was 11% relative to the non-defective specimens.

Confined specimen fracturing was due to fracturing of the CFRP. When fracturing occurred, a thin layer of concrete attached to the CFRP, indicating, therefore, good bonding performance between the two materials. Fractures occurred outside CFRP overlap areas.

We suggest that the ACI equation for estimating confined concrete strength is conservative, as no relationship could be deduced regarding unconfined concrete strength and the deviations between estimated and experimental values.

An issue that should be taken into account in evaluating reinforced structures is that the reinforced specimens were less rigid than the non-reinforced specimens for stress values representing 40% of the concrete ultimate strength value.