Strengthening of reinforced concrete beams using ultra high performance fibre reinforced concrete (UHPFRC)

Introduction

              A novel technique used to improve the performance of existing structural elements is the application of additional Ultra High Performance Fibre Reinforced Concrete (UHPFRC) layers or jackets in connection to the existing elements. The efficiency of this technique has not been adequately studied, and there are not any published studies on the evaluation of this method with comparisons to other traditional strengthening methods such as the use of Reinforced Concrete (RC) layers and jackets. The technique of strengthening using additional RC layers and jackets is one of the most commonly used techniques in seismic areas. There are several published experimental and theoretical studies on beams and columns strengthened with conventional concrete. A crucial parameter in this technique, which can considerably affect the durability and the performance of the strengthened structures, is the concrete shrinkage strain of the additional layers/jackets. Additional stresses are induced in strengthened elements, and cracking of the new layer and/or debonding may occur. The use of UHPFRC could potentially improve both durability and resistance due to its superior mechanical properties.

Reinforced concrete beams prior to strengthening: Numerical modelling and  experimental validation

            The Initial, prior to strengthening, Beam (IB) examined in this study is based on a previous experimental program. Initial beam’s cross sectional dimensions were 150 mm by 250 mm and the length was equal to 2200 mm. The reinforcement consisted of two bars with a diameter of 12 mm (2H12) made of steel with a characteristic yielding stress value of 500 MPa in the tensile side with a cover of 25 mm. The characteristic cylinder concrete compressive strength of the initial beam at 28 days was found equal to 39.5 MPa. The effective span was equal to 2000 mm and the beam was tested under a four-point bending loading with an imposed deflection rate of 0.008 mm/s. The distance between the two loading points in the middle of the span was equal to 500 mm. For the finite element analysis, ATENA software was used. Concrete was simulated with an eight-node element, with nonlinear behaviour and softening branches in both tension and compression using SBETA constitutive model. The ascending compressive branch of this model is based on the formula.

recommended by CEB-FIP model code 90, while its softening law is linearly descending from the peak stress until a limit compressive strain, which was defined by the plastic displacement and the band size, using the fictitious compression plane model. In tension, linear ascending branch and exponential softening branch based on the fracture energy needed to create a unit area of a stress free crack were used. In all the analyses smeared crack approach was used. For the simulation of steel bars, linear elements with bilinear behaviour were used. The numerical results (IBnum) are compared to the respective experimental for (IBexp) and the results are presented in. From the results presented in, very good agreement between the numerical and the respective experimental results was observed. The same assumptions were used for the modelling of RC layers and beams presented in the following sections.

3. Experimental investigation and numerical modelling of UHPFRC

3.1. UHPFRC material preparation

         UHPFRC is a material with enhanced strength in tension and compression and significantly high energy absorption in the post-cracking region. One of the main characteristics of UHPFRC is the enhanced homogeneity which is achieved by using fine aggregates only. In the mix design of the present study, silica sand with maximum particle size of 500 lm was used together with silica fume and Ground Granulated Blast Furnace Slag (GGBS). Silica fume, with particle size almost 100 times smaller than cement, improve not only the density of the matrix but also the rheological properties, while GGBS is used as a partial replacement of cement. High steel fibre content (3%) of straight fibres with 13 mm length and 0.16 mm diameter were used. The mix design is presented and it was based on a previous experimental investigation.

            For the preparation of UHPFRC the dry ingredients were mixed first for 3 min in a high shear mixer Zyklos (Pan Mixer ZZ 75 HE), then the water and the superplasticizer were added to the mix and, at the end, the steel fibres were added gradually. The specimens were cured in a steam curing tank at 90 _C for 3 days and the testing was conducted 14 days after casting. These curing conditions were found to be appropriate for the acceleration of the curing, since the strength achieved after 3 days in the steam curing tank was the same with the strength achieved 3 months under normal curing conditions.

3.2. Compressive and direct tensile tests and numerical modelling

           The standard cube compressive tests (100 mm side) were conducted and the mean compressive strength was found equal to 164 MPa while for the tensile strength, direct tensile tests of 6 dog-bone specimens were carried out.

           A constant loading rate of 0.007 mm/s was used to control the tests which is in agreement with the loading rate used by Hassan et al. leading to comparable results. The extension of the specimens was recorded using Linear Variable Differential Transformers (LVDTs). The setup of was used to measure the average extension over a gauge length of 105 mm, and the stress versus strain (extension normalized to the gauge length) results of all the 6 specimens together with the average curve are presented in.

          The experimental results indicate a variation of the tensile strength between 11.74 MPa and 14.20 MPa. An average stress– strain curve was calculated and the average strength was found equal to 12 MPa. The Young’s modulus was experimentally obtained from the slope of the initial linear part of the stress–strain graph in the linear part and a value of 57.5 GPa was calculated.

The setup used for the optical measurement of the crack together with the strain distribution at the moment when the first crack appeared, are presented in, alongside with stress– strain results for one of the examined specimens.

            Digital Image Correlation (DIC) system was used during the testing in order to monitor the crack opening and the strain distribution. According to these results, the strain is uniformly distributed along the specimen in the elastic part of the stress–strain distribution (strain values below 0.001). Then, in the second phase (strain between 0.001 and 0.005) there is a combination of micro cracks and elastic strain in the neck of the dog bone specimen and, in this phase, the multiple micro cracks opening was taken into account as an average crack opening along the monitoring length. For strain values higher than 0.005, all the extension of the specimen was due to the crack opening.

Fig. 5. Stress–strain results for the direct tensile tests

Fig. 4. Direct tensile tests 

(a)\geometry and (b) experimental setup.

3.3. Experimental and numerical investigation of UHPFRC layers under

flexural loading

               In this section, experimental and numerical results of layers with 50 mm depth were examined. During the casting process, the specimens were cast by pouring the material centrally along the long side and filling the first layer up to approximately 90% of the height of the specimen before the compaction by external vibrator. Afterwards, the moulds were filled and compacted as described in BS EN 14651:2005, and then the specimens were rotated over 90_ around their longitudinal axis for testing. Three identical specimens were tested under 4point loading, with 100 mm breadth, span length of 300 mm, and distance between the two loading points 100 mm. Two LVDTs were used to record the deflection of the layers in both sides, and the tests were conducted using a displacement control of 0.001 mm/s. An external yoke was used in order to exclude any additional displacement at the supports. The testing setup is presented in and a typical crack pattern is illustrated.

         The examined layers were modelled using the assumptions presented in Section 3.3. The numerical model is presented in, and the strain contours together with the crack pattern at midspan deflection equal to 10 mm, are presented in. A deflection equal to10 mm was selected as this was the typical mid-span deflection at the end of the tests.

             The examined layers were modelled using the assumptions resented in Section 3.3. The numerical model is presented in Fig. 11a, and the strain contours together with the crack pattern at midspan deflection equal to 10 mm, are presented in Fig. 11b. A deflection equal to 10 mm was selected as this was the typical mid-span deflection at the end of the tests (Fig. 10b).

            The strain localization in the middle of the span observed in the  experimental investigation (Fig. 10b) was in agreement with the numerical simulation’s results presented in Fig. 11b.

           The load deflection predictions are compared with the respective experimental results of all the three specimens [34] and the results are presented in Fig. 12.

           The numerical results are in good agreement with two of the examined layers, while in one of the examined specimens the strength was considerably lower compared to the other two. This could be attributed to local deficiencies due to the fibre distribution in the mix which resulted to a premature failure of the specimen.

           The assumptions presented in Section 2 and in Section 3.3, for the initial beam and the UHPFRC layers respectively, were used for the simulation of the strengthened beams with UHPFRC.

3.5. Experimental investigation of UHPFRC shrinkage

           In case of strengthened elements with concrete jackets, crucial parameters for the response of these ‘composite’ elements are the interface between the old and new concrete, and the shrinkage of the ‘new’ concrete [15]. Shrinkage strain measurements were recorded over the time for UHPFRC with 3% steel fibres and for Ultra High Performance Concrete (UHPC) without fibres. The mix design of Table 1 was used, and standard prisms 75 mm by 75 mmby 280 mmwere cast. The specimens were stored in a room with relative humidity 42% and temperature 20 _C (Fig. 13a) conditions similar to the standard climate (temperature 23 ± 2 _C and relative humidity or RH 50 ± 5) proposed by DIN 50014.

             The shrinkage strain distribution with the time is presented in Fig. 13b. As a starting point for these measurements (Day 0 in Fig. 13b), the third day after casting was used, when the initial curing was completed.

             The results presented in Fig. 13b indicate that the presence of the steel fibres considerably reduces shrinkage strain values and an average reduction of the shrinkage strain with the time of 30% was observed. In case of UHPFRC with 3% steel fibres (Table 1), the shrinkage strain 90 days after casting was measured equal to 565 microstrains. Shrinkage strain values are highly affected by a number of parameters including the mix design, the curing conditions and the geometry of the examined specimens. Yoo et al. used a special setup to simulate the conditions of free UHPFRC shrinkage effects on slabs and a very steep increase of shrinkage at the very early age was observed. Kamen and Kamen et al. Conducted measurements on specimens with similar geometry to those of the current study (prisms and dog-bone shaped specimens) and a distribution similar to the one presented in Fig. 13b was observed. Also, according to Kamen et al. In the first 2 days there is a chemical shrinkage state after the water–cement contact, followed by a swelling, and the main part of shrinkage strain is developed after the first two days.

5. Strengthening with additional RC layers. Comparisons of the two techniques

         In this section, the results of strengthened beams with UHPFRC are compared to respective experimental results of strengthened beams with traditional strengthening techniques using conventional concrete.

         The results of a previous experimental investigation were used. The geometry and the material properties of the initial, prior to the strengthening beam (IB), are presented in Section 2.

      Beams strengthened in the tensile side with RC layer were examined (ST_RCL_TS). In this case, strengthening was performed by adding a new concrete layer of 50 mm thickness in the tensile side with two bars with a diameter of 12 mm (2H12), made of steel with a characteristic yielding stress value of 500 MPa, and concrete cover of 25 mm (Fig. 23a). The 28 days characteristic cylinder concrete compressive strength of the layer was found equal to 39.5 MPa. The reinforcement of the layer was consisted of two bars with a diameter of 12 mm (2H12) with a total amount (volume) of steel exactly the same with the total volume of steel fibres used in

the layer of ST_UHPFR_TS with 3% steel fibres and 12 MPa average tensile strength.

         Beams strengthened in the compressive zone were also examined (ST_RCL_CS) and, in this case, plain concrete layer with 28 days characteristic cylinder compressive strength of 45.4 MPa, and 50 mm thickness was placed on the top of the beams (Fig. 23b). In this case, a loading rate of 0.008 mm/s was used. 

                                     

         This loading rate is equivalent to the one used for the testing of the prisms (0.001 mm/s) presented in Section 3.4. The distance between the load and the support in the full scale elements (750 mm) was almost 8 times higher compared to the respective

distance in the bending test of the prisms (100 mm) and a loading rate 8 times higher was used in order to obtain comparable results.

         The beams were tested under a four-point bending as described in Section 2 and the failure pattern is illustrated in Fig. 24a and b for specimens ST_RCL_TS and ST_RCL_CS respectively.

The load deflection results for specimens ST_RCL_TS and ST_RCL_CS together with the experimental results of the Initial Beam (IBexp) are presented in Fig. 25.

       From the results presented in Fig. 25, the moment at yield and failure for the strengthened specimens were calculated together with the respective increment, in comparison with the Initial Beam’s (IB) results of Table 3 (Table 4). In case of ST_RCL_TS, the yielding moment was defined by the yielding of the reinforcement of the additional layer, which occurs before the yielding of the reinforcement of the initial beam.

       These results were compared to the respective values calculated for the beams strengthened with UHPFRC with 12 MPa ultimate tensile strength, and the comparisons of My and Mu for the various techniques are presented in Fig. 26a and b.

         The results of Fig. 26 indicate that very high moment increments were observed for beams strengthened with an additional RC layer in the tensile side (ST_RCL_TS), since increments of 150% and 97% were observed for My and Mu. The respective increments for specimens strengthened with UHPFRC in the tensile side (ST_UHPFRC_TS) were found to be lower and equal to 29% and 31%. The results of Fig. 26 indicate that the highest moment increment was observed for a three side UHPFRC jacket and this was found equal to 167% and 178% for My and Mu respectively.

         In case of beams strengthened in the compressive zone, application of UHPFRC (ST_UHPFRC_CS) resulted to an increment of 19% and 28% for My and Mu, while the respective increment was slightly lower (25% for My and 22% for Mu) when normal concrete was used (ST_RCL_CS).

      The technique of strengthening in the tensile side with combination of UHPFRC layer and two steel bars was also examined. The same geometry and reinforcement with the one presented in Fig. 23a was used with the only difference that instead of normal concrete, UHPFRC with 12 MPa ultimate tensile stress was used (ST_UHPFRC_TS_12 MPa and steel bars). The load deflection results were compared to the respective results of ST_RCL_TS and with the initial, prior to strengthening, beam (IB) (Fig. 27).

       The results presented in Fig. 27 indicate that the initial stiffness and the ultimate load capacity were increased when conventional concrete was replaced by UHPFRC in reinforced concrete layers applied to the tensile side. The increment of the yield and ultimate bending moment values are presented in Table 5.

      The results of Table 5 indicate that there is an increment of 7% in the yield bending moment and 9% in the ultimate bending moment when the normal concrete of the layer is replaced by UHPFRC. The results indicate that even if the performance is overall enhanced, the contribution of the UHPFRC in this case was not fundamental.

6. Conclusions

        In this study, extensive experimental and numerical investigation was conducted to investigate the efficiency of UHPFRC for the strengthening of existing beams. Mechanical testing and shrinkage strain measurements were conducted in order to determine UHPFRC material properties, and these results were used for the numerical modelling. The following two conclusions were drawn regarding the behaviour of UHPFRC in tension and regarding the shrinkage strain.

1.The proposed stress–strain model in tension, which was consisted of an initial linear elastic part and a tri-linear post elastic behaviour can accurately predict the response of UHPFRC.

2.Shrinkage strain measurements were also presented for UHPFRC with 3% steel fibres and without steel fibres. The shrinkage strain was 30% reduced in case of UHPFRC with 3% steel fibres, compared to the respective measurements of plain UHPC.

            Extensive numerical modelling was conducted on beams strengthened with UHPFRC. Strengthened beams with additional layer in the compressive, in the tensile side, and with a three side jacket were examined. A parametric study with different values of UHPFRC tensile strength was also conducted and the following observations were made.

1.As expected, in case of specimens strengthened with UHPFRC in the compressive side, increment of the tensile strength of UHPFRC was not considerably affecting the response of the strengthened specimens. Ultimate moment increment less than 4% was observed when UHPFRC tensile strength was increased from 8 MPa to 16 MPa.

2.In case of strengthened specimens with UHPFRC in the tensile side, the ultimate moment was increased by 31% when UHPFRC tensile strength was increased from 8 MPa to 16 MPa.

3.The respective increment for strengthened specimens with three side jackets (ST_UHPFRC_3SJ) was significantly higher and equal to 53%.

           The effect of the post-peak (softening) stress–strain behaviour of UHPFRC on the overall performance of the strengthened beams was also investigated.

1.As expected, for beams strengthened in the compressive side, the effect of the post-peak behaviour of the tensile stress–strain model was negligible.

2.In case of beams strengthened in the tensile layer with UHPFRC layer, when the stresses in the post-peak branch were increased by 50%, the ultimate strength and moment were increased by 8%, while, when 50% lower stresses were used, the ultimate load and moment were reduced by 11%.

3.In case of beams strengthened with a three side jacket the respective ultimate strength increment was 2%, for 50% higher post-peak stresses, while reduction equal to 3% was observed for stress–strain model with 50% lower post-peak stresses.

      Sensitivity analysis was also conducted for the examined strengthened with UHPFRC beams, using different shrinkage strain values for UHPFRC. The main conclusion of this study was that in this case the response of the strengthened elements was not affected considerably by variations in the shrinkage strain value of UHPFRC.

Critical comparisons of this novel technique with the traditional method of strengthening with RC layers were also conducted and the following conclusions regarding the efficiency of the two techniques were drawn.

1.The highest moment increment was observed for a three side UHPFRC jacket and this was found to be equal to 167% and 178%, for My and Mu respectively. When three side jacket was used, the slip at the interface was considerably reduced, compared to the respective values of beams strengthened in the compressive or tensile side.

2.The increment in case of specimens strengthened with UHPFRC in the tensile side was found equal to 29% and 31% for My and Mu. Considerably higher increment was observed in case of beams strengthened with an additional RC layer in the tensile side (150% increment of My and 97% increment of Mu).

3.In case of beams strengthened in the compressive zone, the addition of UHPFRC resulted to an increment of My and Mu 29% and 28% respectively, while the increment was slightly lower (25% for My and 22% for Mu) when normal concrete was used, which indicates that there is no need for high strength concretes when strengthening in the compressive zone.

       For beams strengthened in the tensile side, combination of steel bars and UHPFRC was also investigated and an increment of 7% in the yield bending moment and 9% in the ultimate bending moment was observed, compared to the respective values of beam strengthened with RC layers. The results indicate that even if the performance is overall enhanced, the contribution of the UHPFRC in this case was not fundamental.

The main conclusion of this study is that superior performance can be achieved by the use of three sides UHPFRC jackets. In practise, UHPFRC could be used following the same procedure with the one used for RC jackets. The application of UHPFRC could be done using formworks by adapting the rheological properties of UHPFRC [26]. This novel technique has a great potential for the structural upgrade of the existing structures.

Reference

1.http://fulltext.study/preview/pdf/265998.pdf