Engineering Structures 33 (2011) 20252043

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Strengthening of flat slabs with post-tensioning using anchorages by bonding

Duarte M.V. Faria , Vlter J.G. Lcio, A. Pinho Ramos
Department of Civil Engineering, Faculdade de Cincias e Tecnologia, Universidade Nova de Lisboa, 2829-516 Caparica, Portugal

article

info

abstract
This work aims to study a new flat slab strengthening technique based on post-tensioning with anchorages by bonding using an epoxy adhesive. The main advantages of this technique over the traditional prestress strengthening systems that use mechanical anchorages are that it does not need external permanent anchorages, meaning that the forces are introduced into the concrete gradually instead of being localized, thereby preserving aesthetics and useable space. The seven tested slab models show that this technique meets its objective as it is able to reduce reinforcement strains at service loads by up to 80% if the strengthening technique is applied in two directions and slab deformations by up to 70%, consequently making crack widths smaller. It can also increase punching load capacity by as much as 51% when compared to non-strengthened slabs. The results are compared with the EC2 (2004) [20], ACI 318-08 (2008) [23] and MC2010 (2010) [21] provisions. The main conclusions are that this strengthening technique is effective regarding ultimate and serviceability states and that it represents an advance in RC slab strengthening techniques. 2011 Elsevier Ltd. All rights reserved.

Article history: Received 23 September 2010 Received in revised form 25 February 2011 Accepted 28 February 2011 Available online 29 March 2011 Keywords: Post-tensioning Strengthening Flat slabs Punching

1. Introduction The most common strengthening techniques used in slabs are related to increasing punching and/or flexural capacity and deformation control. The most widely used are the introduction of additional longitudinal reinforcement, with or without a section increase [1], strengthening by means of epoxy-bonded steel plates [24] or fibre reinforced polymers (FRP) [57], strengthening by replacing concrete with a higher grade concrete or fibre reinforced concretes [8,9], strengthening using concrete collars [10] or steel collars [8,10] and strengthening by introducing new shear reinforcements [8,1014]. These techniques are known as passive since the strengthening system is only mobilized when new deformations appear. Active techniques reduce the existing deformations, cracking and stresses caused by bending and punching. Little research has been carried out on these techniques where steel strands are used, although they have already been used in practical applications [15,16]. More recently the use of prestressed FRP has become more popular as studies are being developed. The traditional active technique using prestressing steel strands and external permanent anchorages allows strengthening to flexure and punching simultaneously; deformation and cracking

behaviour also improves. However, it also has some disadvantages that must be taken into account when deciding which technique to use. The technique described here sets out to eliminate some of the disadvantages of the traditional prestressing techniques. This paper describes the experimental research conducted on a new reinforced concrete slab strengthening technique and presents the results obtained. This strengthening system consists of introducing post-tensioning using anchorages formed by bonding a prestressing steel strand to the concrete, using an epoxy adhesive agent for the purpose. Compared with the traditional strengthening using external prestressing, this technique does not need external permanent anchorages; it does not compromise aesthetics and useable space and, whereas in the traditional technique the anchorage forces are localized, in this system the anchorage forces are introduced gradually through bonding. This represents new knowledge and developments in the field of RC slab strengthening. 2. The system 2.1. Construction stages The strengthening system proposed here consists of introducing post-tensioning using anchorages formed by bonding a prestressing steel strand and the concrete. The strengthening procedure is based on the following stages (Fig. 1): drilling the slab (Fig. 1(a)) and setting up the strands (Fig. 1(b)), prestressing the

Corresponding author. Tel.: +351 962821685; fax: +351 212 948 398. E-mail addresses: duamvf@gmail.com (D.M.V. Faria), vlucio@fct.unl.pt (V.J.G. Lcio), ampr@fct.unl.pt (A.P. Ramos). 0141-0296/$ see front matter 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.engstruct.2011.02.039

2026

D.M.V. Faria et al. / Engineering Structures 33 (2011) 20252043

Nomenclature Suffices i Individual 0 At the stage before transfer of prestress initial and ini Initial, after transfer final and fin Final, at maximum load in test y and z Directions of reinforcement Notation

cp max trans
d fccm fck fcm ft fy h P u P

c V Vdev Veff Vexp VRm Es FtransB FtB FbB

Average normal concrete stresses Reinforcement ratio of bar reinforcement Prestress load variation Bond stress Average value for maximum bond stress Average value for allowable transmission bond stress Average effective depth Concrete compression strength on 150 150 150 mm3 cubes characteristic concrete compression strength on 150 300 mm2 cylinders concrete compression strength measured on 150 300 mm2 cylinders Ultimate strength of reinforcement Yield strength of reinforcement Slab depth Prestress/strand load Length of the perimeter control (u = c + 4 d in EC2, u = c + 4d in ACI 318-08, u = c + d in MC2010) Column side dimension Load applied to the slab Vertical component of prestress forces crossing the control perimeter Effective punching load Experimental punching load Mean value of punching resistance Modulus of elasticity of steel Transmitted force at B Force at the top in B Force at the base in B

strut and a deviator, positioned at the top of the slab. Only the deviator stays in the structure and thus must be embedded in the slab finishing. Steel struts, steel mechanical actuators and steel deviators were developed and built as described below. The steel strut was designed to be used in different lengths and to adjust to several prestress profiles. The strut is divided into two sections, one of which may be inserted into the other; it is adjusted by means of a thread and two screws used as a set. Fig. 2 illustrates the equipment. As Fig. 2 shows, there are two parts in each end of the strut that connect to the mechanical actuators, described below. These parts give the strands the desired slope and are connected to the strut with bolts, for easier assembly The actuators are also shown in Fig. 2. These mechanical actuators are activated using a wrench and three bolts. This mechanical system allows the prestress forces to be maintained without loss while the epoxy adhesive is cured. The deviators give the appropriate curvature to the strand and are positioned above the slab, as shown in Fig. 2. 2.3. General considerations As mentioned before this system allows strengthening to flexure and punching simultaneously and promotes an improvement regarding deformation and cracking behaviour. Normally these problems exist in slabs with relatively high slenderness and/or lack of reinforcement, or slabs with poor quality concrete due to construction and/or design errors. Relating to its installation we may say that two men took in average 2.5 h to drill the holes, assemble the system including the prestressing of the strands and injecting the bonding agent. As the steel struts are divided in two parts it is not difficult to lift them up, making its installation easy. 3. Background considerations The bond between the steel strands and the concrete is important to this strengthening technique. An experimental programme of pull-out and push-in tests was developed to study it. These two types of test are designed to simulate the bond behaviour that may be found in the present strengthening technique. Pull-out tests simulate the behaviour of a strand when its tension is increased by loading on the slab; push-in tests simulate the behaviour of the strand when the prestress force that is applied before injecting the hole with the bonding agent (the bonding agent used was HILTIs HIT-RE 500) is transferred to the concrete by bonding. This experimental programme and its results have already been presented [1719], and so, in this manuscript only a synopsis of the tests and its results is presented for better understating of the following developments. Pull-out tests consist in pulling out the strands sealed with the bonding agent in a concrete block (Fig. 3). Five tests were performed for each embedment length, which were 100, 150 and 200 mm long. The experimental results from some pull-out tests are presented in Fig. 4. These figures show the relationship between the pull-out force and slip. The push-in tests consisted in drilling a hole through a concrete block from one side to another and then inserting a high strength steel strand. Afterwards the strand was tensioned, with the help of a mechanical actuator, that allows the load to be maintained while the adhesive is injected and cured. After curing of the bonding agent the strands were de-stressed in the base end, and the load difference between both sides was borne by bond (Fig. 5). Afterwards the strand was pulled-out from the top side. Forces in each end were measured with load cells and slip was measured with the help of four displacement transducers diametrically opposed (two in each end of the strand). In each test, forces and

steel with temporary anchorages (Fig. 1(c)), injecting with a bonding agent (Fig. 1(d)), releasing the provisional anchorages and transferring the prestress forces to the concrete (Fig. 1(e)). Although the system represented in Fig. 1 is unidirectional, it can be bidirectional and have several strands on each column side, as long as certain restrictions (mentioned below) are respected regarding the prestress forces for effective punching calculation, which are limited by geometrical considerations (Section 6.2). If the problem is in the roof slab then steel strands may be positioned above the column. Deviators are only supported near the centre, above/close to the column and act as cantilevers. 2.2. Equipment Specific equipment is needed to apply the prestress. Most of this equipment is not required once the bonding agent has been cured and it can be reused in other prestressing operations. This equipment consists of a strut capable of sustaining the horizontal component of the prestress force, two actuators at the ends of the

D.M.V. Faria et al. / Engineering Structures 33 (2011) 20252043

2027

Fig. 1. System description.

Fig. 2. Equipment used to apply the prestress, (left) adjustable strut, (top right) assembled mechanical actuator and (bottom right) deviators.

displacements were measured continuously, making it possible to determine the forces and slips in each side (base and top), at any time. In the Fig. 6 the load evolution at both ends is shown. After the strand insertion in the hole, it was stressed until it reached the desired load (in this case about 170 kN), represented by point A. At this point the load at the top and in the base are equal. The epoxy adhesive was then injected and cured. At this stage the tension in the strand was reduced at the base end and bond stresses developed along the embedment length. As the base force decreases, it reaches a point (point B) where the top force also starts to drop. Point B stands for the maximum transmitted force (FtransB = FtB FbB ) that a determined embedment length is able to transmit by bonding without losses at the top end. This is the force to keep in mind regarding the maximum force that may be installed in the strands used in the strengthening technique described before, varying with the effective embedment length available in the slab.

From these tests it was possible to obtain for several bonded lengths, maximum pull-out and transmittable loads. These experimental results were compared with theoretical results regarding maximum pull-out and transmittable loads and also draw-in results. Theoretical results were obtained by solving the governing equation of the bond phenomenon adopting a nonlinear local bond/slip law derived from pull-out tests with a short embedment length. Based on the results it is concluded that it is reasonable to assume an average constant bond stress in both cases, making it more user-friendly, since maximum pullout loads and transmittable loads increase approximately linearly with the increase of the embedment length hef . Therefore, the determined average values for pull-out and push-in bond stresses used to determine maximum pull-out and transmittable loads are of 12.0 MPa and 5.2 MPa, respectively. These tests make it possible to quantify the maximum allowable initial force to install in the strand, taking into account the drilled length available and the

2028

D.M.V. Faria et al. / Engineering Structures 33 (2011) 20252043

Fig. 3. Pull-out test.

75.0 62.5 50.0 Load (kN) 37.5 25.0 12.5 0.0 0 5

PO-H1-100

10
PO-H3-100

15
PO-H4-100

20

25 Slip (mm)

30

Fig. 4. Pull-out test results.

maximum force that can be supported by the strand anchorage. These forces are obtained by the product of the respective average bond stress, trans or max , with the bonded area considering the strand diameter. In [19] were also provided design values for trans and max .

Long-term tests were also performed. These tests were very similar to push-in tests and consisted in measuring the loss force in bonded strands for 16 months. Tests showed that the average losses were of 13% and that most of the losses take place in the first two months [19].

D.M.V. Faria et al. / Engineering Structures 33 (2011) 20252043

2029

Fig. 5. Push-in test.

200.0 175.0 Base Force (kN) 150.0 125.0 100.0 75.0 50.0 25.0 0.0 0.0 25.0 50.0 75.0 100.0 125.0 150.0 175.0 200.0 B A

Top Force (kN)

Fig. 6. Push-in test results.

2030

D.M.V. Faria et al. / Engineering Structures 33 (2011) 20252043

TOP VIEW

Fig. 7. Test geometry.

4. Experimental work 4.1. Models and tests The experimental work consisted of testing seven slabs to failure by punching. These slabs measured 2300 2300 mm2 ; slabs DF1DF3 were 100 mm thick, and slabs DF4DF7 were 120 mm thick. Slabs DF1 and DF4 were reference slabs for comparison purposes, while all others were strengthened. Slab DF7 was strengthened bidirectionally while the others were only strengthened unidirectionally. The punching load was applied by a hydraulic jack positioned under the slab, via a central 200 200 mm2 steel plate. Eight points on the top of the slab were connected to the strong floor of the laboratory using strands and spreader beams. Fig. 7 shows a plan and elevation of the test arrangement, including the strands drawn for slab DF7, the only specimen with strands in both directions. Fig. 8 is a photo of the same slab. The models simulated the area near a column of an interior slab panel up to the zero moment lines. The bottom reinforcement of the slab consisted of 6 mm rebars every 200 mm, in both orthogonal directions. In slabs DF1DF3 the top reinforcement consisted of 10 mm rebars every 60 mm and in slabs DF4DF7 10 mm rebars every 75 mm were used, in both orthogonal directions. The slab thickness (h), average effective depths (d) and top flexural reinforcement ratios ( ) are presented in Table 1. In

Fig. 8. Slab DF7 top view.

the strengthened slabs, except DF7, the prestress direction was the direction of the rebars at smaller effective depth. Tests were intended to simulate slabs with relatively high slenderness, since these are the kind of slabs prone to be strengthened with this technique and in laboratory it was chosen to use models that were able to reproduce such slenderness. This study was not directed to study size effect since EC2 [20] and MC2010 [21] present different

D.M.V. Faria et al. / Engineering Structures 33 (2011) 20252043

2031

Fig. 9. Detailed prestress profile, for slabs DF2 and DF3. Table 1 Geometric and material properties. Model ha (mm) db (mm) c (%) fccm (MPa) fcm (MPa) DF1 100 69 1.91 31.0 24.8 DF2 100 67 1.97 33.0 26.4 537 541 561 537 DF3 100 67 1.97 31.5 25.2 DF4 120 88 1.20 24.7 19.8 DF5 120 85 1.24 26.0 20.8 656 637 678 648 DF6 120 84 1.26 26.3 21.0 DF7 120 89 1.19 27.0 21.6

6 10
a b c

fy (MPa) ft (MPa) fy (MPa) ft (MPa)

Model depth. Average effective depth of bar reinforcement. Average reinforcement ratio.

approaches to study punching behaviour, where size effect is taken into account, and it may be used in the present study. Concerning reinforcement ratios, our study indented to examine some slabs with a high reinforcement ratio, since slabs with relatively high slenderness are commonly constructed with a higher reinforcement ratio, thus simulating more accurately a real case. The strands were positioned 50 mm from the column sides and a radius of 2500 mm was adopted for the deviator, and therefore for the strands. The slope of the strands in the embedded portion was about 1/5 (11.3), as shown in Fig. 9. In the experimental tests the deviators were just seated on the slabs and were not covered by a screed. Nevertheless, in an real strengthening situation, deviators must be covered, but this screed (slab finishing) main purpose is to provide protection to deviators and strands and not to provide bond. After the holes were drilled, surveys of the real hole lengths and positioning were carried out to determine the real slopes and lengths available for bonding. These values are important to compute the effective vertical prestress forces and the effective bonded length. It is important to mention that the holes were drilled from the top of the models, as would be done in practice. Analysing these measurements it was found that the average length of the damaged concrete on the bottom of the slab due to drilling (bottom end in Fig. 9) is about 103 mm while on the top end it is about 170 mm, meaning that there is a considerable reduction in relation to the complete hole length of about 500 mm. Damage on the top face of the slab was greater because the vibration induced when the drilling begins is higher than when drilling reaches the bottom of the slab. Regarding damage caused by drilling, both damages (at the top and at the bottom) are important, since in none of them the strand is completely involved/confined and surrounded by the bonding agent. The holes were 18 mm in diameter and were made using an electro-pneumatic rotary impact drill with carbide tipped bits. These holes were afterwards cleaned by blowing air through and brushing. The strands were cleaned of rust, grease and dust,

inserted into the holes and tensioned using the temporary struts and mechanical actuators. This tensioning was done gradually and from both ends, in order to avoid different forces in both strand ends. After this, the epoxy adhesive was injected into the concrete hole from bottom to top, avoiding the formation of air bubbles. The injection process followed the procedure described in [18,19], for the push-in tests. It used an injection system that mixes an epoxy resin with a hardener in a mixing nozzle. After the mixing an exothermic reaction takes place to form a polymer matrix, which becomes the bonding agent. The reference slabs DF1 and DF4 were tested by applying a vertical load in the centre of the slab. In the strengthened slabs, before the prestress operation and while the bonding agent was curing, the vertical load was kept constant at about 40% of the punching failure load of the respective reference slabs. Only after the release of the provisional equipment (Fig. 1) was the vertical load increased up to slab failure. All the failures were by punching. 4.2. Monitoring Four load cells were used (Fig. 7) to measure the vertical load applied to the models, one on each spreader beam. Up to nine displacement transducers were used to measure the vertical displacement of the top surface of the slabs during the tests. In model DF1 only the displacement transducers 15 were used and the displacement transducers 8 and 9 were only used in model DF7. Strain gauges were glued to the top reinforcement of the models (Fig. 10) and were placed perpendicular to the strengthening direction in strengthened slabs, except in DF7, to correspond to the rebars at greater effective depth. The strains in each monitored rebar were measured using a pair of diametrically opposed strain gauges in order to eliminate deviations in the measurements, i.e. those due to flexure of the rebars. The strains were computed from the average of the values obtained in each pair of strain gauges. The forces in each prestressing strand were measured with a load cell during the prestressing operation while using the temporary strut (Fig. 7) and with a pair of strain gauges for each strand during the loading of the slab. During the prestress operation it was possible to correlate the force measured in the load cells with the strains measured in the strain gauges. After the release of the provisional anchorages and transfer of the prestress forces to the concrete (Fig. 1(e)), the forces in the strands were computed using the relation obtained in the first stage. Fig. 10 shows the positioning of the strain gauges and of the displacement transducers. 4.3. Materials properties Compression tests on cubes of 150 150 150 mm3 (fccm ) were carried out on the same day as the test of the corresponding

2032

D.M.V. Faria et al. / Engineering Structures 33 (2011) 20252043

STRAIN GAUGES

Fig. 10. Displacement transducers (top left), reinforcing bars strain gauges (top right), and strands strain gauges (bottom).

slab. The results are listed in Table 1, together with cylinder compression strengths (fcm ) calculated as 0.80fccm . The yield and ultimate strengths of the reinforcement are also included. The strands used were 15.2 mm diameter with a 139.5 mm2 cross section, having 273 kN of maximum force and 246 kN of proof force at 0.1%, and a modulus of elasticity of 197.4 GPa. The epoxy adhesive was also tested for its flexural and compressive resistance. Briefly the results from flexural and compressive testing showed that the average elastic tensile strength was 49.1 MPa and the average yield compressive strength was 108.8 MPa. 5. Discussion of the results Punching is characterized by brittle behaviour at failure with a sudden drop in resistance just after the peak load and it was registered in all slabs. Fig. 11 shows photos of several models after punching failure. 5.1. Load/evolution of strand forces As explained in Section 4.2 a load cell was placed at one end of each strand. A pair of strain gauges was also glued to each strand (Fig. 10) and so it was possible to compute the load in each strand during tests. The force/strain measurement relation was approximately linear within the test range, making it easy to correlate the strain measurement with the strand load during the tests. While the vertical load was kept constant at about 40% of the punching failure load of the respective reference slabs, the prestress was applied to the slab. Initial prestress forces were

determined based on previous research results relating to pushin tests and on the effective bonded length (Section 4.1), applying the uniform bond model with the average bond stresses referred to earlier trans (Section 3). From the observation of Table 2 it may be seen that maximum initial bond stress (i,0 ) was only slightly higher that the average value of 5.2 MPa (Section 3) in some cases. Fig. 12 presents the relationship between the strains and the strand forces during prestressing (on the left) and those between the strand forces and the applied loads in the tests to failure (on the right). The later graphs start from the situation when the strands were locked off against the temporary frames, i.e. that at the end of the graphs on the left. At this point, there is an instant loss of force in the strands which is associated with the elastic shortening of the slab, since at transfer of prestress by bonding to the slab, the horizontal component of prestress force induces compression stresses in the slab, causing slab shortening. This elastic shortening implies also a reduction of the prestress forces. Compression stresses were computed based on the scheme presented in Fig. 13. Compression stresses are computed based on an average width obtained with the width between assumed points of prestress force application in the middle of the bonded length (lprestress appl. ) and in a width crossing the column axe (lslab axe ) considering a force degradation at 45. Applying this methodology an average instant prestress loss of 13 kN is computed, that is very close to the average registered loss of 12 kN, although with some scatter. After this operation, the slab is loaded until failure and the evolution of strand forces is approximately linear. Table 2 presents the results for strand loads and the corresponding installed bond stresses in each stage. It also shows prestress forces losses and gains, respectively, for the transmission and loading stages. The load increase until failure in each strand averages 28.8 kN, corresponding to a strand stress increase of about

D.M.V. Faria et al. / Engineering Structures 33 (2011) 20252043

2033

Fig. 11. Models appearance after punching failure. Table 2 Strand loads and bond stresses for different test stages. Model DF2 DF3 DF5 DF6 Strands S1* S2 S1 S2 S1 S2 S1 S2 S1 S2 S3 S4 Pi,0 a (kN) 64.8 62.1 55.0 63.0 73.6 75.8 64.2 74.9 66.9 65.2 75.7 77.2

i,0 b (MPa)
4.86 4.48 5.66 5.63 4.86 4.48 5.66 5.63 4.86 4.48 5.66 5.63

Pi,initial c (kN) 48.0 40.5 50.5 52.0 58.3 59.4 53.9 74.3 55.3 51.5 46.4 33.0

i,initial d (MPa)
3.60 2.93 5.20 4.65 4.58 4.05 4.76 4.11 4.53 4.33 3.67 2.53

Pi,initial e (kN) 16.8 21.6 4.5 11.0 15.3 16.4 10.3 0.6 11.6 13.7 29.3 44.2

Pi,final f (kN) 84.3 76.7 78.1 78.6 76.1 97.6 79.0 96.7 67.2 62.7 33.3 33.0

i,final g (MPa)
6.32 5.54 8.04 7.03 5.97 6.66 6.98 5.34 5.72 5.46 2.40 2.30

Pf ,final h (kN) 36.3 36.3 27.6 26.6 17.8 38.2 25.1 22.4 11.8 11.2 13.1 0.0

DF7

a b c d e f g h *

Initial prestress load in each strand. Bond stress corresponding to Pi,0 . Prestress load after transmission in each strand. Bond stress corresponding to Pi,initial . Instant prestress load loss in each strand = Pi,0 Pi,initial . Prestress load at punching failures in each strand. Bond stress corresponding to Pi,final . Prestress load increase until punching failure in each strand = Pi,final Pi,initial . S stands for strand.

2034

D.M.V. Faria et al. / Engineering Structures 33 (2011) 20252043

Fig. 12. Force/strain relationships for strands during prestressing (left) and slab load/strand force relationships during tests (right).

205 MPa, not taking into account the strands from model DF7, since there was a problem during the bonding agent injection, that compromised the intended bond behavior (Fig. 12). Also, bond stresses (i,final ) in this stage (punching failure) are far from the maximum value obtained in pull-out tests of 12.0 MPa (Section 3). 5.2. Load/displacement results This section presents the results obtained from the analysis of vertical displacements of the slabs. The displacements are

presented as averages of pairs of values measured by symmetrical displacement transducers, in relation to the centre of the slabs (transducer D3 Fig. 10). Fig. 14 shows displacements for slabs DF1 and DF4 measured at the several displacement transducers. In the graphs, for example, D1 and D5 represent the average of the values registered by displacement transducers 1 and 5, relative to D3. For slab DF1 between a load of about 5 kN to about 50 kN the development of flexural crack occurs and beyond this load cracking stabilizes. For model DF4 these loads are about 20 kN70 kN, respectively.

D.M.V. Faria et al. / Engineering Structures 33 (2011) 20252043

2035

Fig. 12. (continued)

Fig. 13. Scheme of load degradation and widths for compressive stress calculation.

These results are mainly to be compared with the results obtained in the corresponding strengthened models. Fig. 15 shows the evolution of the displacements of strengthened slab models DF2, DF3, and DF5DF7. A considerable decrease in displacements may be observed at loads of about 80 kN when the prestress was applied. This decrease is more visible in the prestress direction, corresponding to the alignment of D1 with D5, than in the orthogonal direction, except in DF7 where the prestress in both directions affected both directions displacements. As an example, in DF2 the maximum displacement fell by about 35% after the prestress was transmitted to the slab. After the strengthening there was a slight increase in slab stiffness caused by the prestress application. It is also important to note that when the prestress is applied with the help of the temporary steel strut the force application point is close to the struts ends, while when the prestress is transferred to the slab and the temporary strut is removed, the application point of that force moves to the interior of the slab. Some of the initial decrease of displacements is lost, as may be observed in Fig. 15.

Fig. 14. Displacement evolution with load for models DF1 and DF4.

Displacement reduction and consequent crack width reduction is considerable for all models, meaning that this system is effective regarding serviceability limit states. Table 3 presents the maximum displacements measured for different load steps (V ), in all models. Table 4 presents the prestress forces and corresponding

2036

D.M.V. Faria et al. / Engineering Structures 33 (2011) 20252043

Fig. 15. Displacement evolution with load for models DF2, DF3, DF5DF7. Table 3 Maximum displacements (mm). Model V = 150 kN (1) DF1 DF2 DF3 DF4 DF5 DF6 DF7 9.5 4.6 4.1 6.3 2.8 2.1 1.7 (2) 4.7 5.1 5.2 3.7 3.3 1.9 V = 180 kN (1) 12.1 6.4 5.5 8.4 4.2 3.4 2.4 (2) 6.3 6.6 7.2 5.2 4.9 2.5 V = 210 kN (1) 8.1 7.6 5.7 5.2 3.6 (2) 8.0 8.4 6.8 6.7 4.0 V = 240 kN (1) 10.0 9.7 7.4 6.9 5.2 (2) 9.7 10.5 8.6 8.4 5.6 V = 270 kN (1) 12.3 9.4 8.9 6.9 (2) 11.9 10.7 10.5 7.4 V = 290 kN (1) 11.1 10.5 8.1 (2) 12.6 12.2 8.6 V = 315 kN (1) 10.1 (2) 10.7

(1) Direction parallel to strands. (2) Direction perpendicular to strands.

deviation forces after the transmission stage (initial) and at punching failure (final). Vdev,ini is computed based on the initial prestress load (Pinitial = Pi,initial from Table 2) and on the measurements of the real slope of the drilled holes (Section 4.1). Since the models deformed during the tests, leading to larger vertical deviation of the

strands and to an increase in their load (at punching failure designated as Pfinal = Pi,final from Table 2), the deviation forces are also expected to increase. As the displacements of the slab and the evolution of forces in the strands were controlled during the tests, it was possible to compute the final vertical deviation forces (Vdev,fin ).

D.M.V. Faria et al. / Engineering Structures 33 (2011) 20252043 Table 4 Initial prestress load and corresponding deviation forces. Model DF2 DF3 DF5 DF6 DF7
a b c d

2037

Pinitial a (kN) 88.40 102.50 117.70 128.20 186.16

Pfinal b (kN) 160.98 156.70 173.70 175.74 196.13

Variation (%) 82.1 48.9 47.6 37.1 5.4

Vdev,ini c (kN) 30.60 37.90 49.90 57.40 74.50

Vdev,fin d (kN) 61.30 63.10 79.60 84.40 84.90

Variation (%) 100 66.5 59.5 47.0 14.0

Initial prestress load. Final prestress load. Initial deviation forces. Final deviation forces.

effects were more visible, since there was also prestress in the direction of the instrumented rebars. In this model a considerable decrease in rebar strains can be seen, leading to smaller strain values at punching failure than in the other models. This decrease reaches about 80% when compared with slab DF4 for load steps of 150 and 180 kN. 6. Punching loads 6.1. Punching load capacity All the slabs failed by punching and their ultimate loads (Vexp ), including self-weight, are given in Table 5. An analysis showing the effect of prestress on the punching capacity is presented in this section. To make allowance for the variations of slab depths, ratios of bar reinforcement and concrete strength, the experimental ultimate loads were divided by resistances calculated by EC2s [20] expression for the characteristic strength of ordinary rc slabs, but with the characteristic concrete strengths replaced by the actual mean strengths (Table 1) and with the limit on the depth factor k ignoredsee Eq. (1). The resulting ratios of strength were divided by the ratios for slabs DF1 (for the 100 mm slabs) and DF4 (for the 120 mm slabs) to isolate the effects of the strengthening. The results obtained are shown in Fig. 18. It is clear that the strengthening was effective with average increases of punching load being 40% for slabs DF2 and DF3 as compared with DF1 and 51% for DF5DF7 when compared to DF4. 6.2. Comparison of experimental results and code provisions
Fig. 16. Strain evolution with load for models DF1 and DF4.

Comparing model DF1 with models DF2 and DF3, the average decrease in displacements is about 50%, for load steps of 150 and 180 kN. Comparing models DF5 and DF6 with model DF4, that reduction is about 55% in the prestress direction and 30% in the other direction. For model DF7 these reductions average about 70% and 65%, respectively, compared with model DF4. 5.3. Load/strain results This section presents the results obtained from the analysis of strains in the longitudinal reinforcing bars of the slabs. The strains presented below are the averages from the pairs of gauges shown in Fig. 10. The yield strain of the main bars was approximately 2.7 . Fig. 16 shows that in the reference slabs only the steel bar positioned closest to the column starts to yield at a load of about 155 kN in model DF1. All strengthened models registered a decrease of strains during the prestress operation (Fig. 17). This decrease would have been bigger if the strain gauges had been positioned in the same direction as the prestress direction. In model DF7 the prestress

The resistance without punching shear reinforcement, using EC2 [20] was calculated with the following expression (Eqs. (1)(4)) : VRm = 0.18 k (100 fcm )1/3 + k1 cp u d

(1) (2) (3) (4)

= y z 0.02 cp =
d=

cpy + cpz
2 2

dy + dz

The limitation of the parameter k = 1 + 200/d in EC2 [20] to a maximum of 2 was neglected and k1 = 0.1. In the quantification of punching resistance the mean values for the compressive resistance of concrete were used and the partial safety coefficient was neglected. Reinforcement ratio values are calculated taking into account a slab width equal to the column width plus 3d for each side. Deviation forces are computed based on the work of Ramos [22], who proposed that their calculation should be based on the vertical components of prestress forces in the strands running within distances of 0.5dp from the column sides (dp is the prestress strand effective depth). In Fig. 19 is presented pictures of saw cuts

2038

D.M.V. Faria et al. / Engineering Structures 33 (2011) 20252043

Fig. 17. Strain evolution with load for models DF2, DF3, DF5DF7.

1.70 1.60 1.50 1.40 1.30 1.20 1.10 1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00

surface approximately intersects the base of the deviators. All other strengthened slabs exhibited a similar behaviour. For the calculation of the resistance without punching shear reinforcement and without any prestress effect, using ACI 318-08 [23], the relevant expression for square columns, with side lengths less than 4d is Eq. (5): VRm = 4 fcm u d 12

(5)

In prestressed slabs the following expression is used (Eq. (6)): VRm =

0.29 fcm + 0.3 cp u d + Vdev .

(6)

Fig. 18. Relative punching strengths for slabs DF13 and slabs DF47.

of slabs in both orthogonal directions. It is possible to observe that the punching failure angle is higher in a cut perpendicular to the strengthening direction. Actually, the punching failure

According to ACI 318-08 [23], Eq. (6) is applicable when fck 35 MPa (corresponding to an approximate value of fcm 43 MPa), with bidirectional prestress and when the average compressive stress in concrete in each direction due to prestress is between 0.9 and 3.5 MPa. Though this expression is not strictly applicable to most of the models presented here, it was nevertheless used to

D.M.V. Faria et al. / Engineering Structures 33 (2011) 20252043 Table 5 Comparison between effective experimental loads and code provisions taking into account the initial deviation forces. Model DF1 DF2 DF3 DF4 DF5 DF6 DF7
a b c d e *

2039

Pinitial a (kN) 0.00 88.40 102.50 0.00 117.70 128.20 186.16

Vdev,ini b (kN) 0.00 30.60 37.90 0.00 49.90 57.40 74.50

Vexp c (kN) 190.72 272.94 254.64 199.00 295.00 292.72 319.52

Code EC2* ACI 318* EC2* ACI 318* EC2* ACI 318* EC2* ACI 318* EC2* ACI 318* EC2* ACI 318* EC2* ACI 318*

Veff d (kN) 190.72 190.72 242.34 272.94 216.74 254.64 199.00 199.00 245.10 295.00 235.32 292.72 245.02 319.52

VRm e (kN) 202.58 137.79 204.82 156.01 202.22 161.55 217.41 167.49 219.15 202.50 217.97 209.31 235.02 245.18

Veff /VRm 0.94 1.38 1.18 1.75 1.07 1.58 0.92 1.18 1.12 1.46 1.08 1.40 1.04 1.30

Initial prestress load. Initial deviation forces. Experimental punching load. Veff effective punching load: EC2 Veff = Vexp Vdev,ini ; ACI 318-08 Veff = Vexp . Predicted failure load. Taking into account the assumption described in Section 6.

estimate the ACI 318-08 [23] predicted punching load. Deviation forces are computed based on the vertical component of prestress forces crossing the control perimeter defined in ACI 318-08 [23]. In both codes cp was computed based on Fig. 13, considering the width at the face of the column along the perpendicular direction of the strands and on the slab depth. Table 5 gives a comparison between experimental loads and the predicted values obtained using EC2 [20] and ACI 318-08 [23], taking into account the initial deviation forces. For the reference slabs DF1 and DF4, EC2 [20] leads to expected mean punching resistances very close to those obtained experimentally, with an average ratio Veff /VRm of 0.93, slightly against safety. ACI 318-08 [23] provides more conservative results, since the experimental punching loads reached between 18% and 38% above the average prediction, with an average Veff /VRm of 1.28. With regard to the strengthened models EC2 [20] gives an average Veff /VRm of 1.10 and ACI 318-08 [23] the ratio is 1.50. So, both EC2 [20] and ACI 318-08 [23] is somewhat conservative when using the initial deviation forces. Table 6 presents a comparison between experimental loads and the predicted values obtained using EC2 [20] and ACI 318-08 [23], taking into account the final deviation forces (Section 5.2 and Table 4). Using this methodology, mean punching resistances are closer to the experimental value, leading to a reduction of conservatism in both EC2 [20] and ACI 318-08 [23]. With respect to the strengthened models, the mean Veff /VRm for EC2 [20] is reduced to 0.96 (with a COV of 0.04) and for ACI 318-08 [23] it becomes about 1.29 (with a COV of 0.07). Regarding EC2 results it is important to it was ignored the limitation of the remind that parameter k = 1 + 200/d to a maximum of 2 and so all the results are on the safe side. Recently, it was published the first draft of MC2010 [21]. Punching design recommendations in MC2010 [21] present a new design philosophy based on the critical shear crack theory described in [24,25], for slabs without and with transverse reinforcement, respectively. In MC2010 [21] the design expressions are presented, while in [24] are presented the expressions for the average values Eq. (7) that can be compared with the experimental results. VRm ud

be obtained by Eq. (8).

= 1.5

rs fy d Es

msd mrd

1.5

(8)

If the slab is prestressed may be obtained using Eq. (9):

= 1.5

rs fy d Es

msd mPd mRd mPd

1.5
(9)

fcm

3/4 1 + 15 d +d g g0
d

(7)

where is the slab rotation, dg is the maximum aggregate size and dg 0 is a reference size equal to 16 mm. The rotation of the slab may

where msd is the average moment per unit length for calculation of the flexural reinforcement in the support strip and can be approximated for inner columns without moment transfer as V /8 (V is the punching load); mRd is the design average flexural strength per unit length in the support strip; mPd is the average decompression moment in the support strip due to prestressing and rs indicates the position where the radial bending moment is zero with respect to the column axis. Combining Eq. (7) that describes the failure criterion with Eq. (8) or Eq. (9) that describes the loaddeflection behaviour and setting VRm equal to V , it is possible to obtain iteratively the punching strength. In Table 7 is presented the results for VRm for each tested slab. MC2010 [21] states that deviation forces from strands applied inside the basic control perimeter could be deducted to the column reaction. In Table 7 is presented values for VRm according to Eq. (7), using Eq. (8) (without considering the prestress effect in the slab rotation) and using Eq. (9) with the initial and final prestress forces (from Table 4), since MC2010 [21] mentions that punching load should be computed based on the maximum slab rotation of both orthogonal directions and since most slabs are strengthened in one direction. This comparison was also performed in slab DF7. Experimental values of slab rotation were obtained using maximum displacements of the slab (Figs. 14 and 15) assuming that this rotation concentrates near the column faces. From the observation of Table 7 is possible to verify that the better approximation of the computed VRm with the experimental effective punching load is obtained when using Eq. (9) with final prestress forces and deviation angle, as had happened with other codes. Table 7 shows that using Eq. (9) when computing VRm of prestressed slabs gives better results than using Eq. (8), although most slabs are only strengthened in one direction. In Table 8 is presented a comparison between values for d (corresponding to the VRm values presented in Table 7) computed

2040

D.M.V. Faria et al. / Engineering Structures 33 (2011) 20252043

Fig. 19. Slabs saw cut in both directions.

using Eq. (8) (without considering the prestress effect in the slab rotation) and applying Eq. (9) using initial and final prestress forces with experimental values. From the obtained results it is possible to find that values calculated for d using Eq. (9) are closer to the registered ones for the strengthened models. The introduction of deviation forces originated by the strands relieve the effective punching load near the column, as these act in opposite direction of the gravity loads. Additionally, there is an introduction of compression stresses that causes a decrease in crack opening, proportional to d presented in Table 8. By its observation is possible to see that those measured values are smaller in the prestressed direction, and that in slab DF7 crack opening is almost the same in both directions. Nevertheless, comparing for example slab DF1 with slabs DF2 and DF3, it is seen that although these last slabs are only prestressed in one direction, crack opening in the other direction also decreases (assuming that slab rotation of slab DF1 is similar in both directions) meaning that prestress has also an influence in the non-prestressed direction. Regarding DF4 it would be expectable that crack opening was

higher when compared to slabs DF5DF7, and this may be due to experimental results scatter also noticed by Muttoni [24]. Observing results presented in Table 3 and Figs. 14 and 15, it was registered a decrease in displacement of strengthened slabs in relation to their reference slabs for the same load levels. Concerning steel strains and observing Figs. 16 and 17 it is possible to see a decrease in steel strains in all strengthened slabs with a considerable decrease in slab DF7. In Fig. 20 is presented the relation between normalized Veff punching load ud and crack opening d. It is clear that there f
cm

is a tendency of a decrease in normalized punching load with crack opening. In this figure is also clear that DF4 results are slightly out of the range of the remaining ones. In all codes comparison, strands were not considered in the calculation of flexural reinforcement ratio since these are not bonded in the punching area. The bonded portion of the strands is positioned far from the punching failure surface and with a slope. As deviators are only supported near the centre and act as

D.M.V. Faria et al. / Engineering Structures 33 (2011) 20252043 Table 6 Comparison between effective experimental loads and code provisions taking into account the final deviation forces and strands final forces. Model DF1 DF2 DF3 DF4 DF5 DF6 DF7
a b c d e *

2041

Pfinal,pun a (kN) 0.00 160.98 156.70 0.00 173.70 175.74 196.13

Vdev,fin b (kN) 0.00 61.30 63.10 0.00 79.60 84.40 84.90

Vexp c (kN) 190.72 272.94 254.64 199.00 295.00 292.72 319.52

Code EC2* ACI 318* EC2* ACI 318* EC2* ACI 318* EC2* ACI 318* EC2* ACI 318* EC2* ACI 318* EC2* ACI 318*

Veff d (kN) 190.72 190.72 211.64 272.94 191.54 254.64 199.00 199.00 215.40 295.00 208.32 292.72 234.62 319.52

VRm e (kN) 202.58 137.79 207.43 191.80 204.02 190.27 217.41 167.49 221.58 236.63 219.99 240.01 235.49 256.41

Veff /VRm 0.94 1.38 1.02 1.42 0.94 1.34 0.92 1.18 0.97 1.25 0.95 1.22 1.00 1.25

Final prestress load. Final deviation forces. Experimental punching load. Veff effective punching load: EC2 Veff = Vexp Vdev,fin ; ACI 318-08 Veff = Vexp . Predicted failure load. Taking into account the assumption described in Section 6.

Table 7 Comparison between effective experimental loads and MC2010 code provisions taking into account the initial and final strand deviation and compression forces. Model DF1 DF2 DF3 DF4 DF5 DF6 DF7 Veff (kN) using Vdev,ini 190.72 242.34 216.74 199.00 245.10 235.32 245.02 Veff (kN) using Vdev,fin 190.72 211.64 191.54 199.00 215.40 208.32 234.62 VRm (kN) using Eq. (8) 156.41 154.77 151.85 180.57 176.45 174.68 189.34 Average COV
a b c

Veff /VRm 1.22 1.57 1.43 1.10 1.39 1.35 1.29 1.40a 0.07a 1.22 1.37 1.26 1.10 1.22 1.19 1.24 1.26a 0.05a

VRm (kN) using Eq. (9), with initial prestress 156.41 160.94 159.76 180.57 186.45 186.40 204.96 Average COV

Veff /VRm 1.22 1.51 1.36 1.10 1.31 1.26 1.20 1.28b 0.10b

VRm (kN) using Eq. (9), with final prestress 156.41 168.20 165.83 180.57 193.70 193.22 207.61 Average COV

Veff /VRm 1.22 1.26 1.16 1.10 1.11 1.08 1.13 1.15c 0.16c

Values are for strengthened slabs only; for slabs DF1 and DF4 the average relation is 1.16 with a COV of 0.07. Values for all slabs; if only strengthened slabs are considered the average relation obtained is of 1.33 with a COV of 0.09. Values for all slabs; if only strengthened slabs are considered the average relation obtained is of 1.15 with a COV of 0.06.

Table 8 Comparison between the experimental and computed product d of the slab computed based on MC2010 expressions for . Model Experimental d (mm) Strands 1 and 2 direction DF1 DF2 DF3 DF4 DF5 DF6 DF7 1.33 1.17 1.10 1.27 1.45 1.33 1.47 Opposite direction 1.18 1.15 1.14 1.63 1.53 1.44 1.44 1.46 1.45 1.60 1.62 1.62 1.64 Average COV 0.92 0.80 0.76 0.79 0.90 0.82 0.89 0.83a 0.07a 0.81 0.79 0.71 1.01 0.94 0.88 0.88a 0.10a 1.44 1.32 1.27 1.60 1.42 1.39 1.35 Average COV 0.88 0.86 1.03 0.96 1.09 0.96 0.10 0.89 0.90 1.15 1.10 1.07 1.02 0.12 1.44 1.18 1.15 1.60 1.28 1.26 1.31 Average COV 0.92 0.99 0.96 0.79 1.13 1.05 1.12 1.05b 0.07b 1.00 1.00 0.71 1.27 1.21 1.10 1.12b 0.11b

d using Eq. (8)

Exp./Computed

d using Eq. (9) with

initial prestress

Exp./Computed

d Eq. (9) with

final prestress

Exp./Computed

a b

Values are for strengthened slabs only. For slabs DF1 and DF4 the average relation is 0.81 with a COV of 0.11. Values are for strengthened slabs only. If all slabs are considered the average relation is 1.02 with a COV of 0.15.

cantilevers, only the slope of the strands in the straight portion is used in the calculation of deviation forces. 7. Design guidelines Regarding the design of a slab strengthened with the system proposed, and having in account the results of code comparison presented in the last section, both EC2 [20] and MC2010 [21] may be used. If the initial prestress and deviation forces are used both codes provide safe estimates of the punching load and this is the suggested philosophy of design. Nevertheless it is possible

to estimate the strand force increment taking into account the slab rotation. As seen before, MC2010 [21] provides an expression for the calculation of (Eqs. (8) and (9)) as a function of slab punching load, and assuming that this is the slab rotation of the slab that takes place between the transmission phase failure (this is close to reality as in most cases prestress will be chosen to reverse almost all if not all existing slab deformations), it is possible to estimate strand force increment and therefore final compression and deviation forces from prestress. The increase of strands length ( l) in the unbonded portion of the strand, for each side of the column, can be assumed to be proportional to the

2042

D.M.V. Faria et al. / Engineering Structures 33 (2011) 20252043

0.65

0.60

0.55

0.50

0.45 1.00

1.20

1.40

1.60

1.80

Fig. 20. Normalized punching loads versus crack opening.

presents the results obtained. This active strengthening technique enables the simultaneous solving of the several deficiencies with respect to the traditional technique of external prestressing with external anchorages. Compared with the traditional strengthening using external prestressing, this method does not require external anchorages thus improving the aesthetics of the repaired slab and maintaining the clear headroom below it. An experimental programme was developed to study the efficiency of the proposed strengthening technique when used to strengthen flat slabs. This technique reduced the slabs deflections at service loads up to 70% as compared with unstrengthened slabs, and reduced crack widths significantly. The experimental research carried out also shows that there was a decrease in the average strains of the reinforcing bars, mainly in the bidirectional strengthened slab. The load capacity of the strengthened slabs was higher than the reference slabs. The increments in load capacity varied between 36% and 54%. The calculation of the vertical deviation forces above the columns, using the initial deviation forces led to punching resistances lower than those obtained experimentally. When the final vertical deviation forces were used, conservatism was diminished. Both EC2 [20] and MC2010 [21] provided a better agreement between the predicted and the experimental results, whereas ACI 318-08 [23] was somewhat conservative. Regarding all said before it may be stated that the experimental results indicate that the system is effective with respect to serviceability and ultimate behaviour. It was given guidelines for the correct estimation of maximum installed force in the strands due to slab rotation. Acknowledgements

Fig. 21. Predicted and measured load increase versus strand length increment.

slab rotation , taking into account the strands effective depth dp and the depth of the compression zone in the concrete cross section ( l (dp x)). This increase in strands length is composed of two portions: one that is the elastic elongation of the strand and the other that is the slip of the bonded portion of strand for a determined strand force. The slip of the bonded portion is dependent of the strand force (P > Pi,initial ), on the initial transmitted force (Pi,initial ), on the bonded length and on the bond behaviour, and may be computed based on the results and analysis presented by Faria et al. [19] and resumed in Section 3 of this manuscript. The elastic elongation may be computed by the elastic theory and is dependent on the unbonded length of the strand. A comparison of the experimental results and the computed values is presented in Fig. 21 where is presented the theoretical relation between strand load and the increase of strands length l for the several strengthened slabs and also the experimental results (where for l = 0, P = Pi,initial ). It may be observed that the theoretical results are always higher that the experimental results. This may be ascribed to the fact that the theoretical analysis does not take into account that the strand effective depth varies along its development towards the centre of the slab. Using the computed values presented in Table 8 using Eq. (9), that are inferior to the experimental ones, the theoretical values for force increase would actually be closer to the experimental ones. By doing so it is possible to estimate maximum prestress force installed in the strand and thereby compare it to the maximum allowed force limited by the maximum bond capacity and yielding of the strand. 8. Conclusions This paper describes the experimental research conducted on a new reinforced concrete slab strengthening technique and

This work received support from the Fundao para a Cincia e Tecnologia-Ministrio da Cincia, Tecnologia e Ensino Superior through scholarship number SFRH/BD/37538/2007 and Project PTDC/ECM/114492/2009. We would like to thank Concremat for making the slab models and HILTI Portugal who supplied the epoxy adhesive (HILTI HIT-RE 500) and all the equipment related to adhesives and drilling. References
[1] Datta AR, Seraj SM. Effect of overlay on punching shear capacity of slabs. J Civ Eng Inst Eng Bangladesh 2003;CE 31(2). [2] Zhang JW, Teng JG, Wong YL, Lu ZT. Behavior of two-way RC slabs externally bonded with steel plate. ASCE J Stuct Eng 2001;3907. [3] Van Gemert D. Special design aspects of adhesive bonding of plates. Farmington Hills (MI, USA): ACI Special Publication SP 165 (2), Repair and Strengthening of Concrete Members with Adhesive bonded Plates, American Concrete Institute, 1996, p. 2541. [4] Ebead U, Marzouk H. Strengthening of two-way slabs using steel plates. ACI Struct J 2002;99(1):2331. [5] Teng JG, Chen JF, Smith ST, Lam L. FRP strengthened RC structures. England: John Wiley & Sons; 2002. [6] Seim W, Horman M, Karbhari V, Seible F. External FRP poststrengthening of scaled concrete slabs. ASCE J Compos Constr 2001;(May):6775. [7] Binici B, Bayrak O. Upgrading of slabcolumn connections using fibre reinforced polymers. Eng Struct 2005;27(1):97107. [8] Ramos AMP, Lcio V, Regan PE. Repair and strengthening methods of flat slabs for punching. In: International workshop on punching shear capacity of RC flat slabs. Royal Institute of Technology. Department of Structural Engineering. 2000. [9] Harajli MH, Maalouf D, Khatib H. Effect of fibers on the punching shear strength of slabcolumn connections. Cement Concr Compos 1995;17(2):16170. [10] Hassanzadeh G, Sundqvist H. Strengthening of bridge slabs on columns. Nordic Concr Res 1998;21. [11] Ghali A, Sargious MA, Huizer A. Vertical prestressing of flat plats around columns. Detroit: ACI SP 42- Shear in Reinforced Concrete, 1974 p. 90520. [12] Kunz J, Ruiz MF, Muttoni A. Enhanced safety with post-installed punching shear reinforcement. In: Tailor made concrete structures-proceedings of the international fib symposium. 2008. p. 67984.

D.M.V. Faria et al. / Engineering Structures 33 (2011) 20252043 [13] Ruiz MF, Muttoni A, Kunz J. Strengthening of flat slabs against punching shear using post-installed shear reinforcement. ACI Struct J 2010;107(4):43442. [14] Duarte I, Ramos A, Lcio V. Strengthening of existing flat slabs with transverse reinforcement. In: Proceedings of CCC 2008, challenges for civil construction. FEUP. 2008. [15] Provenghi SG, Brandow GE. Evaluation, repair and reinforcement of an ultrathin two-way flat slab with drop panels. SP 128-79. ACI. p. 127492. [16] Bondy KB. Externally applied post-tensioning systems. Struct Mag 1995;147. [17] Faria DMV, Lcio VJG, Ramos AMP. Strengthening of reinforced concrete flat slabs using post-tensioning with anchorages by bonding. In: fib Symposium concrete: 21st Century Superhero. 2009. [18] Faria DMV, Lcio VJG, Ramos AMP. Bond behaviour of prestress steel strands bonded with an epoxy adhesive and a cement grout for flat slab strengthening purposesexperimental study. In: 3rd fib International Congress. 2010.

2043

[19] Faria DMV, Lcio VJG, Ramos AMP. Pull-out and push-in tests of bonded steel strands. Magazine of Concrete Research. MACR-D-10-00068. September 2010 [in press]. [20] European Committee for Standardization. EN 1992-1-1 Eurocode 2: design of concrete structurespart 1-1: general rules and rules for buildings. 2004. [21] Federation International du Beton. Model Code 2010. First Complete Draft. fib Bulletins No. 55 and 56. 2010. [22] Ramos A. Punching in prestressed concrete flat slabs. Ph.D. thesis. Lisbon: Technical University of Lisbon; 2003. [23] American Concrete Institute: ACI 318-08. Building code requirements structural concrete and commentary. ACI Committee 318. 2008. [24] Muttoni A. Punching shear of reinforced concrete slabs without transverse reinforcement. ACI Struct J 2008;105(4):44050. [25] Ruiz MF, Muttoni A. Punching shear of reinforced concrete slabs with transverse reinforcement. ACI Struct J 2009;106(4):48594.