Concrete Slab Strengthening with CFRP Textile Reinforced Shotcrete

Transcription

1 Concrete Slab Strengthening with CFRP Textile Reinforced Shotcrete Daia ZWICKY Professor, PhD, MSc (ETH) University of Applied Sciences Western Switzerland Daia Zwicky, born 1971, is currently vice-chair of WC4 Conservation of Structures and chair of the Swiss National Group of IABSE. His research interests cover modeling of structural materials, hybrid structural elements, and structural analysis and strengthening of existing structures. Summary This paper reports on a new strengthening system for concrete elements, being composed of carbonfiber reinforced polymer fabrics (CFRP fabrics) embedded in shotcrete. The strengthening system was evaluated in a test series on full-scale single span slab strips of 6 m length, on a non-strengthened reference specimen and two slab strips with different strengthening ratios. The paper discusses the test results with regard to serviceability limit state (flexural stiffness, crack widths, and admissible loads for given deflection limits), and to ultimate limit state (strengthening ratio, behavior at rupture, post-peak bearing capacity, and ductility). It concludes with some analytical considerations on the structural behavior and bearing capacity of the new strengthening system. Keywords: strengthening, reinforced concrete, slab, textile reinforcement, carbon-fiber reinforced polymer (CFRP), shotcrete, full-scale tests, SLS, ULS, bending, anchorage. 1. Introduction Strengthening of existing structural elements in reinforced concrete with textile-reinforced concrete (TRC) is a relatively new strengthening technique. In comparison to strengthening with fiber-reinforced laminates glued with epoxy on the tension face (externally bonded reinforcement, EBR) or glued into slots (near-surface mounted, NSM-EBR) which have become state-of-the-art [1], TRC strengthening techniques have several advantages: Lower requirements on the evenness of the support (= surface to be strengthened) since the application of the first layer of shotcrete also serves as leveling layer. Explicit demand of a moist support for bonding the TRC layer to the strengthened element. For EBR, the moisture in the support has to be considerably lower eventually demanding for expensive drying installations, in particular in humid environment (e.g. tunnel, parking). Lower required qualification level of the staff since the applied glue is well-known. No particular requirements have to be considered in its preparation, the pot life is considerably longer, and the reinforcement textiles can be applied fresh-on-fresh. Improved moisture balance since the adhesive for the textile reinforcement consists of a mineral porous matrix, avoiding vapor sealing on strengthened surfaces. This is not the case for epoxy-based adhesives applied on large surfaces. Increased fire resistance, imminent to the strengthening system, due to the shotcrete cover on the textile reinforcement if the CFRP textile reinforcement is protected against carbon oxidation. This fire resistance is associated with an increase of the allowable strengthening ratio because the strengthening will not fail in the case of fire. Increased corrosion protection of the internal reinforcement thanks to the additional cover. However, TRC strengthening techniques have to compete with traditional strengthening by EBR in terms of structural performance.

2 2. Full-scale tests on slab strips strengthened with TRC system ARMO Experimental results on a new TRC strengthening system called ARMO were gained in a study [2] at the Structures Laboratory of the College of Engineering and Architecture Fribourg (EIA-FR), a member of the University of Applied Sciences Western Switzerland, for the benefit of a Swiss SME, S&P Clever Reinforcement. Additional analytical considerations are amended in section Test specimens Full-scale bending tests up to failure were carried out on single-span reinforced concrete slab strips with a span of L = 6 m, and a cross-section of 85 cm width and 22 cm height (slenderness 27.3). A non-strengthened slab strip served as reference specimen for two further slab strips, reinforced with one and two layers of CFRP fabrics in shotcrete layers of 1 cm and 2 cm thickness, respectively. The geometrical reinforcement ratio in the flexural tension zone amounted to 0.43% (6 Ø12 mm) while 0.20% (4 Ø10 mm) were provided at the flexural compression face. The concrete cover on the transverse reinforcement (Ø8, s = 150 mm) on both faces was approx. 25 mm. The specimens thus represent a typical building slab with usual structural detailing. The specimens were concreted upside down to facilitate the application of the strengthening layers in the laboratory. The supporting face for the strengthening was roughened by rake to a depth of approx. 5 mm immediately after concreting to simulate hydro-jetting usually applied in practice. After hardening, the superficial cement slurry was eliminated. Fig. 1 Uniaxial carbon fiber fabric (ARMOmesh ). The recently developed CFRP fabrics ARMOmesh used in the tests are characterized by statically desired, but relatively expensive carbon fiber rovings in the longitudinal direction which are held in position transversely by cheaper non-structural glass fiber rovings (Fig. 1). The textiles are embedded in a layer of sprayed mortar ARMOcrete employable by dry mix or wet mix method. For both strengthened specimens, the TRC strengthening layer ended 20 cm away from the bearing axis. 2.2 Structural materials The compressive strength of the concrete with a maximum aggregate size of 16 mm was determined for each test specimen by compression tests on three cylinder specimens (Ø160 mm x 320 mm) and varied between 32.1 MPa and 37.5 MPa. The same specimens could be used to determine Young s moduli, varying between 39.8 GPa and 41.4 GPa. The flexural tensile strength was derived from four-point bending tests on prism specimens 150 mm x 150 mm x 600 mm, and yielded 3.3 MPa on the average. Overall, the concrete corresponds to a usual building construction quality. With a total of six tension tests on rebar specimens Ø12 of 650 mm length, average values of the yield strength of 540 MPa, of the tensile strength of 617 MPa, and of the ultimate strain of 56 were found. Due to the significant variation of almost 20% in ultimate strain, the requirements for normal ductility reinforcing steel could not be met, being unfortunately also representative for practical ductility properties. Table 1 Properties of ARMOmesh L500 [3]. Roving layout 2x 1600 tex/roving 58.5 rovings/m Area weight of C-fibers in principal direction g/m 2 Young s modulus (theoretical) 240 GPa Young s modulus (design) 160 GPa Tensile strength of C-fibers (theoretical) 4'000 MPa Tensile strength of roving (experimental) 2750 MPa C-fibers cross-section per fabric layer 105 mm 2 /m Ultimate strain (theoretical) 16.7 The mechanical properties of the CFRP fabrics were not tested separately. Table 1 shows the values provided by the manufacturer. The fabrics are protected from carbon oxidation at elevated temperatures by amorphous silica coating. The mechanical properties of the mortar ARMOcrete, sprayed by the wet mix method, were determined on three cylinder specimens Ø160 mm x 320 mm, providing average values of Young s modulus of 25.2 GPa and of compressive

3 strength of 30 MPa, respectively. These values are intentionally lower than for usual TRC shotcrete since the system wants to increase economic competitiveness by using a cheaper sprayed mortar. 2.3 Test set-up and procedure The test setup approximately simulated a uniformly distributed load by applying point loads at 1/5 of the span, i.e. at 1.2 m. The LVDT measurements covered loads, deflections as well as tension and compression strains in the zone of maximum bending moment. In a first phase, the slab strips were loaded in load steps of 8 kn up to a total load of 32 kn (expected serviceability limit). At this load level, 100 load cycles down to 8 kn were carried out to gain some insight on the structural behavior under repeated loading. After the load cycles, the specimens were tested to failure in 8 kn steps. At every load stage, crack patterns were recorded and maximum crack widths were measured. 3. Evaluation of full-scale test results Fig. 2 shows the deformation behavior up to failure of the tested specimens. The graph only shows the applied jack forces and the associated deflections at mid-span. The dead load of the slab strips can be estimated as 4.68 kn/m while the additional weight of the loading equipment amounts approx. 1 kn per loading point. The self-weight of the strengthening layers amounts to 0.28 kn/m for test D1 and 0.37 kn/m for test D2, respectively. 3.1 Structural behavior at serviceability limit state (SLS) Bending stiffness at cracked state The strengthened slab strips D1 and D2 showed a somewhat stiffer behavior, implying a certain contribution of the TRC layers at cracked state. However, the form of the deflection curves indicates that the textile reinforcements only absorb forces with progressive crack formation. This delayed activation of the TRC layers can be attributed, on the one hand, to an initially required Fig. 2 Load-deflection curves for all tested slab specimens (load cycles at 8 kn hidden).

4 straightening of the passively applied CFRP fabrics. On the other, the bond between textile reinforcement and shotcrete is rather poor, leading to a structural behavior of the fabrics that is closer to unbonded than to perfectly bonded reinforcement (also see section 4.4). Stiffness losses due to repeated loading slightly increase with increased strengthening, amounting to 14% (D1), 17% (D1) and 22% (D2), respectively. Thus, a certain destruction of the bond between CFRP fabrics and shotcrete due to repeated loading can be reckoned. This assumption can be explained by considering the changes in bond stiffness ratios of steel and strengthening textiles. The stronger the TRC strengthening, the more forces it attracts due to the increased stiffness for normal forces. This, in turn, leads to increased bond stresses for the CFRP reinforcement, resulting in accelerated bond destruction and increased loss of bond stiffness. Thus, the repeated loading essentially releases the bond of the strengthening textile, leading to a structural behavior of the latter that is closer to unbonded tendons with end anchorages than perfectly bonded reinforcement SLS loads for given deflection criteria Table 2 shows SLS loads for usual deflection criteria as reached in the tests. The provided values do not consider the stiffness losses due to repeated loads (section 3.1.1). Table 2 Service loads for given deflection criteria. Test load Q ser,exp for a deflection at mid-span of L/500 L/350 L/300 D0 9.4 kn 100% 12.8 kn 100% 14.3 kn 100% D kn 122% 13.2 kn 103% 15.4 kn 107% D kn 138% 16.0 kn 125% 17.5 kn 122% The strengthening layers are more efficient for low load levels and a higher increase in admissible deflection can be assumed for stronger strengthening layers. However, the stiffness increases due to the TRC layers is not significant enough for eliminating SLS problems, i.e. for increasing bending stiffness by applying TRC strengthening layers. 3.2 Behavior at ultimate limit state (ULS) Yielding of the internal steel reinforcement From the moment when the internal reinforcement attains yielding, the CFRP fabrics absorb virtually all additional internal tension forces. The steel only assumes forces at the ratio of its hardening module (< 1% of Young s modulus). Yielding of the reinforcement therefore essentially marks the start of the effective load-bearing of the textile reinforcement. Experimental loads Q y at yielding and associated deflections w y at mid-span can be estimated from the load-deflection curves (Fig. 2 and Table 3). A slight increase of yielding loads is associated with increasing textile reinforcement cross-section implying that the fabrics are marginally activated before yielding. Note that deflections w y at yielding remain practically constant for all specimens. Table 3 Test results at yielding of internal steel reinforcement and at rupture. Q y [kn] w y [mm] Q u [kn] w u [mm] L/w u [-] M u [knm] D s [-] θ pl [mrad] D % % % % (47.2) 100% D % % % % % D % % % % % Behavior at rupture and post-peak bearing capacity Table 3 also shows the attained maximum loads Q u, associated deflections at mid-span w u as well as post-peak bearing capacities Q Rest. Doubling the number of fabric layers doubles the increase in maximum load. The sudden loss in load-bearing capacity of the strengthened specimens (Fig. 2) can be attributed to an excessive slip between CFRP fabrics and shotcrete, i.e. anchorage pull-out failure at the ends of the TRC layers. None of the specimens showed rupture of the internal steel or the textile reinforcement (Fig. 3). Still, the strengthened slab strips exhibited about 10% higher post-peak load-bearing capacity per fabric layer after anchorage failure than the ultimate load of the reference specimen. Together with Q Rest [kn]

5 Fig. 3 Intact carbon grid in largest crack after reaching maximum load. the small load increases (Fig. 2) after anchorage failure, this indicates that there was still some bond strength available between CFRP fabrics and shotcrete. Final rupture of all specimens was attained by softening of the flexural compression zone. The strengthened specimens showed no signs of failure at the interface between TRC layer and concrete support Crack patterns and strains at ultimate Fig. 4 shows the crack patterns of all specimens after failure. In test D0, the measured average crack spacing of ca. 130 mm in the zone of maximum bending moment approximately corresponds unsurprisingly to the spacing of the transverse spacing of 150 mm. In test D1, no decrease in crack spacing could be observed, as the average crack spacing amounted to approx. 150 mm. In test D2, some reduction of average crack spacing could be observed with 110 mm, reflecting a slightly increased contribution of the strengthening layer to tension stiffening. At maximum load, the average strains at the flexural tension face amounted to 11.6 (D0), 10.5 (D1) and 8.6 (D2), respectively, on a length of 1200 mm at mid-span (i.e. between loads Q2 and Q3 in Fig. 4). Average strains at the flexural compression face amounted to -3.7 (D0), -2.5 (D1) and -2.7 (D2), respectively, on a length of 1050 mm at mid-span. As it can be deduced from Fig. 2, all test specimens were initially uncracked. Strains due to self-weight of specimens and loading equipment thus did not considerably influence the strain measurements. D0 D1 D2 Fig. 4 Crack patterns after failure.

6 3.2.4 Ductility The ductility index D s = w u /w y (see Table 3) shows that the plastic deformation capacity decreases to approx. 80% for the strengthened specimens, being independent of the number of strengthening textile layers. An analogous conclusion is drawn if plastic rotation angles θ pl = 4(w u w y )/L are considered, determined from a simple rigid-body failure mechanism on a single span beam [4]. The plastic deformation capacity decrease of the strengthened specimens can principally be expected since the strengthening of the flexural tension face is associated with a higher flexural compression zone. Assuming comparable ultimate concrete strains at the flexural compression face, this increase in compressed height leads to reduced strains at the flexural tension face, implying a reduced curvature in the cross-section and thus, reduced deflections w u at maximum load. 4. Calculation of stresses in the CFRP fabrics 4.1 Experimental references Considering self-weight of specimen and loading equipment (section 3), the maximum bending moments at mid-span amounted to 67.1 knm (D0), 83.3 knm (D1) and 98.3 knm (D2), respectively. At attaining estimated yield forces (Table 3), the associated bending moments at mid-span amounted to 59.5 knm (D0), 63.8 knm (D1) and 66.3 knm (D2), respectively. 4.2 Simplified estimation of fabric stresses The lever arms of inner forces at ULS are estimated as 95% of the associated effective depth d s = 181 mm of the internal reinforcement and d TRC of the textile reinforcement (d TRC = 227 mm for D1 and d TRC = 230 mm for D2), respectively. If steel hardening is neglected, the reinforcing steel contribution is the same for all test specimens and results in a flexural resistance of 63.0 knm, underestimating the experimental value of test D0 by approx. 6% (section 4.1). Consequently, the differences in bending moments at ULS for tests D1 and D2 can be attributed to the CFRP fabrics alone but the associated stresses are somewhat overestimated since steel hardening is neglected. Thus, the stresses in the fabrics at maximum load should have been 760 MPa (D1) and 720 MPa, respectively, being associated with strains of 3.2 (D1) and 3.0 (D2), respectively, assuming a Young s modulus of 240 GPa (Table 1). Applying classical assumptions for cracked concrete until steel yielding (linear-elastic materials, rigid bond between reinforcement and concrete, plane sections remain plane) and neglecting the contribution of compression reinforcement, the flexural compression height can be estimated as approx. 34 mm, being associated with a bending moment of 62.2 knm at steel yielding. This value overestimates the experimental value of test D0 by less than 5%. From the differences in yielding moment, the CFRP stresses can thus be estimated as 200 MPa (D1) and 160 MPa (D2), respectively. 4.3 Detailed derivation of theoretical fabric stresses Methodology The proposal from [5] is amended to account for the hardening of the reinforcing steel, in order to not overestimate the stresses in the strengthening textiles. The underlying assumptions are classical: equilibrium for bending with normal force, plane sections remain plane, rigid bond between reinforcement and concrete, and neglecting concrete tensile strength and compression reinforcement. The reinforcing steel behavior is bi-linearly approximated while the CFRP reinforcement is linearelastic until rupture with a Young s modulus of 240 GPa (Table 1). All calculations are performed with a weighted effective depth of the tensile reinforcements, accounting for the differing stiffnesses of steel and textile reinforcement. The compressed concrete is assumed to be rigid-plastic, allowing the determination of the compression zone height and the associated force by moment equilibrium alone if the applied normal force and bending moment are converted to the common centroid of the internal and strengthening reinforcement. The uniaxial concrete compressive strength is reduced by a factor of (30/f cm ) 1/3 1 [6], amounting here to 0.93 to The reduction factor allows deriving an equivalent rigid-plastic compressive strength and accounts for the more pronounced softening behavior of higher strength concrete [7]. The total

7 tensile force is determined from normal force equilibrium, and the forces in the two reinforcements are derived by considering for relative normal force stiffnesses and lever arms of inner forces. Considering the stiffness change of steel reinforcement at yielding requires some spreadsheet iteration, also allowing the derivation of bending moments at yielding. If the applied bending moment exceeds the yielding moment, the same procedure is applied for the excessive bending moment, but considering the hardening modulus of the reinforcing steel in the determination of the weighted effective depths and the lever arms of inner forces and deducing the compressed height at yielding (since the latter cannot be stressed twice). This approach easily allows determining the additional tensile forces in both, internal and strengthening reinforcement Results and comparison with experiments A bending moment of 64.0 knm is found analytically for test D0 at yielding, implying an acceptable difference of approx. 8% to the experimental value. At ultimate, an analytical strain in the steel reinforcement of 21.6 is found but it cannot be compared directly to the experimental strain of 11.6 (section 3.2.3) since the analysis neglects tension stiffening effects. From the mechanical and geometrical properties of the reinforcing steel and considering an average crack spacing of 130 mm (section 3.2.3), a tension stiffening factor of 0.45 can be determined [8], resulting in an analytical strain of 9.7 that agrees reasonably with the experimental value. For test D1, a bending moment at yielding of 77.3 knm is calculated, overestimating the experimental value by more than 20%. The associated load would already correspond to approx. 90% of the ultimate load and the CFRP textile should have a stress of approx. 630 MPa, considerably exceeding the estimated value of 200 MPa. For the ultimate load, the CFRP fabric stress should amount to 910 MPa, also clearly exceeding the estimated value of 760 MPa (section 4.2). An analytical bending moment of 91.4 knm at yielding is found for test D2, overestimating the experimental value by almost 40% and corresponding to about 90% of the ultimate load. It would be associated with a textile stress of also approx. 630 MPa, lying clearly above the estimated value of 160 MPa. At ultimate load, as stress of approx. 790 MPa should be present in the CFRP fabric that deviates a bit less from the estimated stress of 720 MPa. The comparative calculations above show that the assumption of rigid bond between CFRP fabric and shotcrete is not justified since too high strains (and thus, stresses) are derived in the textile reinforcement. For reference specimen D0, the classical analysis assumptions are justified if tension stiffening effects are accounted for. 4.4 Consideration of CFRP fabrics as unbonded reinforcement The considerable differences in analytical and experimental strains cannot be attributed alone to the fact that the CFRP fabrics cannot be applied completely straight and plane, thus requiring an initial straightening for their stress activation. A considerable strain difference has to be attributed to the poor performance of the bond between textiles and shotcrete. If the textiles are considered as unbonded tension ties without prestressing being anchored at the strengthening layer ends by bond, the strain increase in the CFRP fabrics is rather related to the global deformation behavior than to the strain profile in the cross-section (also see section 3.1.1). For a single span beam, the average textile strain at ULS can be determined approximately from a rigid-body failure mechanism with ε m,trc = 4w u /L (d TRC x)/l cr, where w u = ULS deflection at midspan, L = span, d TRC x = distance of textile reinforcement to neutral axis, and l cr = cracked length between anchorage zones. Note that this estimative approach cannot be applied for determining the textile strain at yielding since a rigid-body failure mechanism should not be assumed at this stage. The distance d TRC x can be determined from the measured average tensile and compressive strains (section 3.2.3), resulting in 181 mm (D1) and 173 mm (D2), respectively. The lengths between anchorages are estimated from Fig. 4 to approx. 4.6 m (D1) and 5.0 (D2). With w u from Table 3, average strains ε m,trc = 3.7 (D1) and ε m,trc = 3.1 (D2) can be calculated, agreeing reasonably for test D1 with the simplified strain estimation of section 4.2 and very well for test D2, and also confirming that the assumption of rigid bond between CFRP fabrics and shotcrete is not justified. It is rather unlikely that the strain reduction factor due to bond would be as low as approx (= ε m,trc /ε m,exp, also section 3.2.3) while the steel reinforcement is yielding [9].

8 Also note that the deflections w u only differ 5% (D1) and 11% (D2), respectively, from the directive [6] which defines ULS of slabs with unbonded prestressing at a maximum deflection of w R = L/ Conclusions and outlook TRC strengthening layers are only activated significantly with attaining internal steel reinforcement yielding, analogously to traditional externally bonded reinforcements EBR. Consequently, TRC strengthening only contributes to ultimate resistance and not to serviceability properties. Considerable degrees of strengthening of approx. 35% per fabric layer could be attained in the tests, but being associated with a ductility reduction of 20 to 30%. The experimentally observed pull-out failure of the fabrics anchorage in the uncracked zones at the ends of the strengthening layer led to a sudden and marked drop in bearing capacity and thus, does not allow exploiting the full CFRP tensile strength. Further experimental campaigns for determining anchorage resistance and associated bond lengths have already been performed and are currently evaluated in the light of developing a design approach. Presuming an appropriate preparation of the support, interface failure between TRC layer and support can principally be excluded. The textile reinforcement of the TRC strengthening system investigated here should rather be regarded as unbonded than bonded reinforcement. The small differences in analytical strains and strains estimated from experimental results can be attributed to the initially necessary straightening of the fabrics that is due to the execution procedure. Analytical investigations can be performed in analogy to external prestressing methods, thus analyzing global deformations of the structure instead of cross-sectional strains. Reaching ULS can be assumed at a maximum deflection of approx. w R = L/40, but the associated strain in the CFRP fabrics should be reduced by a resistance factor of approx. 1.3 [1]. Acknowledgements The study [2] was performed under the guidance of Prof. Dr. René Suter. The effort of all people involved in the tests and providing the results to the author of this paper is highly acknowledged. References [1] SIA 166, Klebebewehrungen (Adhesively Bonded Reinforcement). Swiss Society of Engineers and Architects (SIA), Zurich, Switzerland, [2] SUTER R., EJUPI G., MOREILLON L., Renforcement des dalles en béton au moyen de treillis en fibres de carbone rapport d essais (Strengthening of concrete slabs with CFRP fabrics), College of Engineering and Architecture, CC «matériaux et innovation», Fribourg, Switzerland, Sept [3] > ARMO [4] KENEL A., Biegtragverhalten und Mindestbewehrung von Stahlbauteilen (Flexural behavior and minimum reinforcement of structural concrete elements), PhD thesis, Swiss Federal Institute of Technology, Institute of Structural Engineering, Zurich, Switzerland, [5] JESSE F., CURBACH M., Verstärken mit Textilbeton (Strengthening with textile reinforced concrete), Betonkalender, Ernst und Sohn Verlag, 2010, pp [6] SIA 262, Concrete Structures, Swiss Society of Engineers and Architects (SIA), Zurich, Switzerland, [7] MUTTONI A., Die Anwendbarkeit der Plastizitätstheorie in der Bemessung von Stahlbeton (The applicability of plasticity theory to the design of structural concrete), PhD thesis, Swiss Federal Institute of Technology, Institute of Structural Engineering, Zurich, Switzerland, [8] ZWICKY D. Effects of Construction Details in Existing Concrete Structures on Bond, Bond in Concrete Bond, Anchorage, Detailing, June , Brescia, Italy, p [9] ULAGA T., Betonbauteile mit Stab- und Lamellenbewehrung: Verbund- und Zuggliedmodellierung (Concrete elements with bar and laminate reinforcement: bond and tension chord modeling), PhD thesis, Swiss Federal Institute of Technology, Institute of Structural Engineering, Zurich, Switzerland, 2003.