LONG-TIME CREEP TESTING OF PRE-CRACKED FIBRE REINFORCED CONCRETE BEAMS

Transcription

1 BEFIB2012 Fibre reinforced concrete Joaquim Barros et al. (Eds) UM, Guimarães, 2012 LONG-TIME CREEP TESTING OF PRE-CRACKED FIBRE REINFORCED CONCRETE BEAMS Terje Kanstad * and Giedrius Žirgulis * * COIN, Department of Structural Engineering, The Norwegian University of Science and Technology, Trondheim, Norway terje.kanstad@ntnu.no, giedrius.zirgulis@ntnu.no Keywords: Concrete, synthetic fibres, steel fibres, composite basalt fibres, long-time testing. Summary: This paper describes pre-cracked synthetic and steel fibre reinforced concrete beams tested in five years with constant load. A second series with steel and composite basalt fibre reinforced beams is presently being carried out to confirm the results of the first series. The long-time load level is 50% of the residual strength at 1.5mm crack-width. All the tested beams show similar behaviour, i.e. that during five years the time dependent deflections are in the range 2-3 times the initial deflections, and these results are related to creep coefficient values calculated according to the creep model in Eurocode 2. Furthermore, mapping of the real fibre content is an essential part of the result evaluation, and has been used to explain the different behaviour of similar specimens. One consequence of the present results is that creep due to long time load should be accounted for in crack-width calculations. 1 INTRODUCTION For cracked synthetic fibre reinforced concrete long-time creep and time dependent loss of strength have commonly been considered as major concerns, when the material is used in load carrying concrete structures, and several research activities have been published [5, 6, 7, 8]. To further lighten the problem, and to contribute to the design basis in the service limit states especially, four test rigs were made in 2006[1]. Since then four beams have been loaded for a period of five years, while four more beams have been loaded since September The beams have dimensions b/h/l=120/150/600, are notched in the middle section and are loaded by four points bending. The beams were pre-cracked in a displacement controlled test rig to a crackwidth of approximately 0.2mm, unloaded and finally moved to the long-time test rigs. The long-time test conditions should represent the service limit states, and the load is therefore set to 50% of the residual strength of the reference beams. The rigs have been located in a climate room with temperature 20 o C, and relative humidity equal to 50%. The four beams loaded for five years had the following fibre contents: (1)0.5% end hooked steel fibres, (2) 0.7vol% synthetic macro fibres with embossed surface, and (3, 4) 1.0vol% of the same synthetic macro fibres. Of the four beams in the 2 nd series, 2 beams contained 0.5% end hooked steel fibres, while the 2 remaining contained 1.0% twisted composite basalt fibres. Due to the notch and the pre-cracking process, a hinge mechanism occurs in the middle section of the beams, and the material behaviour in this section is completely decisive for the deflections of the beams. Furthermore can possible explanations of the time dependent deformation be due to the following effects: Creep in the compressive zone of the hinge, time dependent bond failure and bond creep strains between the concrete and the fibres crossing the crack (i.e. creep in the tensile zone), drying shrinkage of the compressive zone, and finally, creep in the fibre materials. In this investigation are creep effects calculated according to Eurocode 2 compared to the experimental values.

2 2 TEST METHODS The test procedure consists of three main steps: (1) Determination of the long-time load level, (2) pre-cracking of the beams in the short-time test rig, and (3) application of constant load in the longtime test rig. 2.1 Bending test for determination of residual strength and pre-cracking The setup and the geometry for the short-time bending tests are shown in Figure 1. The beam dimensions were b/h/l=120/150/600mm, and all beams were prepared with a 30mm deep notch in the middle cross section, which reduced the effective cross section area to 120x120mm 2. The loading rate corresponds to a crack-width opening displacement (CMOD) rate of approximately 0,2mm/min. For the current test setup the residual flexural tensile strength (f ft,eq ) is calculated using the equation: f L F (1) bd ft, eq ( 12) 2 Where F (δ12) is the average value of the load at CMOD 1 =0.5mm and CMOD 2 =2.5mm. It might therefore be argued that F (δ12) represents the residual flexural tensile strength at 1.5mm crack-width. In the pre-cracking test, the beams are loaded until a crack-width of approximately 0.2mm is reached. The purpose of this procedure is to make the long-time test conditions relevant for the service limit states in real concrete structures. Larger crack widths in this part of the test will lead to larger fibre damage than what is typical for service limit states load conditions. 2.2 Long-time test Figure 1: Four point bending test rig The long-time test rigs, shown in Figure 2, are also based on four points bending with the same load location as illustrated in Figure 1. There are four equal test rigs which are placed in a climate room with T=20 C and RH=50%. The notched and pre-cracked beams were placed in the rigs, and the constant load was applied by means of hydraulic cylinders connected to manometers, a system which allows simple control and adjustment of the desired load level. At the first few weeks of testing, the load was adjusted daily or every second day. Gradually the deformation rate of the beams decreased, and therefore the hydraulic pressure stabilised, and longer time intervals between the adjustments were allowed. The beam deformations were measured with a manual extensometer with sufficient accuracy. The samples in the long time test were loaded to a constant load P l corresponding to a residual bending stress of 0,5 f ft,eq., defined as the average of the residual strength at 0.5 and 2.5mm CMOD as described above: 2

3 2 bd Pl 0,5 f ft, eq (2) L This load level can be considered as a typical long-time load for a concrete structure where most of the service limit states loads consist of long time load. However, for most structures, the long time load will be considerably lower than 50% of the ultimate strength. 2.3 Fibre counting Figure 2: Long-time test rig setup To relate the real fibre content and the fibre orientation to the corresponding nominal values is an important part of the result evaluation, and therefore the number of fibres crossing the cracked sections of the tested specimens has been counted. First 50mm wide concrete slices were cut close to the cracks in the middle sections, and the number of fibres determined. Afterwards each slice was crushed into fine pieces so that the fibres could be extracted. The fibres were washed, dried and weighted to find the total amount of fibres within the slices, and finally the real fibre volume fraction in the middle part of the beam was determined. Furthermore the orientation factor was calculated using the following expression: nf Af N Af vf Ac vf Where: n f is the number of fibres per concrete unit surface; A f is the cross-section of a fibre; v f is the fibre volume fraction; N f is the number of fibres in the concrete cross-section; A c is the concrete cross-section area. Under isotropic conditions the orientation factor should be equal to 0,5, while it is 1,0 if all fibres are directed in one direction normal to the considered plane. (3) 3

4 Mix name w/binder Cement CEM I 42,5R, kg Micro silica, kg Water, l Sand 0-8, kg Sand 0-2, kg Crushed 8-16, kg SP, % from cement Fibres volume fr. BEFIB2012: Terje Kanstad, Giedrius Žirgulis 3 TEST SERIES Two test series have been carried out, and the first consists of four beams, one with steel fibres, and three with macro synthetic fibres. These beams were placed and loaded in the climate room in May 2006, and the deflections were measured for more than five years until they were replaced by the second beam series in September The latter series consists of four specimens as well, two beams reinforced with the composite basalt macro fibres and two with steel fibres. The last series is presently running, and is partly carried out to verify the results of the first series. 3.1 Part materials The same part materials have been used to produce the concrete beams in both test series. The natural sand consists of two fractions: 0-2mm and 0-8mm, while the 8-16mm coarse aggregates consist of a crushed rock fraction. The following fibre types were used: (1) 48mm long synthetic (polyolefin) fibres with embossed surface, Young s modulus of about 10 GPa and yield strength of about 600 MPa. (2) 60 mm long hooked end steel fibres with aspect ratio 65, and yield strength above 1000 MPa. (3) Basaltic composite, twisted fibres which are 40mm long, and are having a diameter of 2mm. The failure mechanism of all these fibre types in cracked ordinary strength concrete is fibre pullout, similarly as for most types of steel fibres [4]. The low aspect ratio of the Basalt fibres, which still are in the development phase, are the main explanation of their relatively low contribution to the residual strength. Figure 3 shows photos of the three fibre types. The concrete mix compositions are presented in Table1, and it can be seen that the only deviations between them are due to small differences in cement content. Table 1: Mix composition. The three upper ones are from 2006, while the two lower are from Fibres 0,5 Steel Steel hooked 0, ,25 0,5 (series I) end, 60mm 0,7 Synth 0, ,25 0,7 Plastic straight, 48mm 1,0 Synth 0, ,25 1,0 Plastic straight, 48mm 1% 0, ,32 1,0 Basalt, twisted Basalt 0,5 Steel (series II) 0, ,32 0,5 Steel hooked end, 60mm Figure 3 The synthetic, steel and composite basalt fibres used in the two test series 4

5 3.2 Series I (2006) As previously mentioned, the first test series started in 2006 [1], and the following four beams were used for the long-time test: Two samples (1.0 Synth-1 and 1.0 Synth-3) with 1.0 volume% synthetic fibres; One sample (0.7 Synth-3) with 0.7% of synthetic fibres; One sample (0.5 Steel-1) having 0.5% of steel fibres. Three parallel beams were made for each fibre variant, and the beams not used for the long-time testing (1.0Synth-2, 0.5Steel-1 and 0.7Synth-2) were used to determine the short-time residual strength which formed the basis for determination of the long-time load. After termination of the longtime test, the long time specimens were also tested for residual short-time strength to investigate if these properties are influenced by the long-time load, and other time dependent processes. The beams were produced by casting vertical plates and sawing the beams as illustrated in Figure 4. The loading direction related to the casting process is also shown in the figure. Figure 4: Casting process for the vertical plates, sawing of the beams, and beam loading direction (Series I) 3.3 Series II (2011) For this test series, 8 beams were produced. Four test beams were made with composite basalt fibre reinforced concrete (B1-1, B1-2, B1-3, B1-4) while the remaining four beams were steel fibre reinforced (S0.5-1, S0.5-2, S0.5-3, S0.5-4). The basalt fibre concrete beams contained 1.0 volume % fibres while the steel fibre reinforced beams included 0.5% fibres, the same as one of the beams in Series I. The test beams were of the same size as in test series I, but in contrary to the first series, the beams were cast directly in moulds according to the procedure described in [2], thus avoiding the sawing process. Four beams (two from each type FRC) were tested in the short-time four point bending test, while the remaining four were placed in the long-time test rig. 5

6 4. CALCULATIONS 4.1 Evaluation based on the time dependent creep coefficient The creep coefficient (φ) is usually defined as the ratio between the time dependent creep strain (ε creep ) and the initial or the short-time strain (ε initial ) according to the following relations: creep total initial creep initial 1 where (3) In calculations the creep ratio is introduced in the constitutive equations together with the modulus of elasticity, and the equations can be solved in accordance with the principles of linear visco elasticity for aging materials or simplifications thereof. In the present investigations, the creep coefficient development has been calculated using the model described in Eurocode 2 [3] which is valid for the current cross section dimensions, temperature and relative humidity conditions. As the first step in the evaluation of the present experimental results, the creep coefficient development was calculated from the long-time test results. Because the determination of strains is not straightforward in the current problem, the strains are simply replaced by the corresponding deflections in the expression for the creep ratio: initial creep total initial t => t initial initial l l l l l creep total initial initial initial (4) In which Δl initial are the initial short time deformations, Δl creep the time dependent or the creep deformations while Δl total are the total deformations. 4.2 Evaluation based on stage II analysis with elastic fibre stiffness in the tensile zone Evaluation of the experimental results based on a nonlinear hinge model is presently ongoing. Creep and shrinkage of the compressive zone and creep due to the bond stresses between fibres and concrete in the cracked tensile zone are assumed to be the most important time dependent properties. For the pre-cracked cross section, the fibres are assumed to behave elastically for short-time load, an assumption which is based on pullout testing reported in [4]. The evaluation so far, however, shows that creep in the tensile zone has to be included to explain the relatively large time dependent deformations which occur for all fibre types. 5 RESULTS 5.1 Residual short-time strength, long-time load, and real fibre content According to the test procedure, the short-time residual strength was determined for each beam type before the long-time test could be carried out because the long-time load should be 50% of the short-time residual strength. The latter was determined as the average of the strengths at 0.5 and 2.5mm CMOD. Furthermore was also the short-time residual strength of the long-time test beams determined after the long-time tests had been terminated. All the short-time test results of Series I are presented in Figure 5, and the initial short-time test 6

7 results from 2006 are characterized by the high initial stiffness due to their un-cracked state compared to the cracked long-time test specimens. The general observation from these figures is that long-time loading, as carried out in the present investigation, does not have any significant influence on the residual strength. It is the authors opinion that the deviations between the corresponding curves in Figure 5 mainly can be explained by the varying fibre area in the critical sections, which partly is verified by fibre counting on sawn cross sections as explained in Chapter 2.3. These results are summarized in Table 2 below where the residual strength used as basis for the test program, and the long-time load levels, are reported together with the real fibre volume, the real fibre area crossing the critical sections, and the fibre orientation factors. From the results it is seen that the real fibre area in general is higher than the nominal area, which probably mainly is explained by a favourable fibre orientation compared to isotropic conditions. Comparing the results for the two steel fibre reinforced beams in Figure 5a, it is seen that the beam tested in 2006, before the long time test was started, had a maximum load in the post cracking range on 24kN and a fibre area of 0.30%, while the one tested with long time load for 5 years had a maximum load on 17kN and a fibre area of 0,26%. The corresponding nominal value for the fibre area is 0.25% (=0,5v f for isotropic conditions). The results for the beams with 1% synthetic fibres show that the maximum post cracking load is 13 kn for the beam tested in 2006 while the corresponding fibre area is 0.95%. In this case the corresponding nominal value is 0.5% (=0.5v f ). The beams exposed to long time load, however, showed larger post cracking strength, 14 and 20kN, although the measured fibre area for these are lower, 0.68% and 0.84%, respectively. For the beams with 0.7% synthetic fibres the post cracking loads and real fibre areas are 12.5kN (0.62%) and 14 kn (0.57%). Again the beam exposed to long-time load had the highest capacity. Table 2. Residual strength and load level in the long-time tests. Fibre reinforcement areas. Beam Residual strength Long time load Fibre volume Fibre area Real area/ nominal area Orientation factors 0.5%Steel(I) kN 8.88kN 0.41% 0.26% %Synth kN 5.10kN 0.69% 0.57% %Synth kN 6.13kN 0.99% 0.68% %Synth kN 6.13kN 1.25% 0.84% %Steel(II)-1& kN 8.26kN - 1.0%Basalt-1&2 4.15kN 2.07kN Pre-cracking tests Figure 6 presents typical curves from the pre-cracking tests. It is seen that the crack-widths (CMOD) are limited to about 0,2mm, and that the cracking load in these tests are about the same for all beams. This is logical because the same concrete is used in all the beams and the cracking load is expected to be only slightly influenced by the amount of fibres used in the present investigations. It can also be seen that the load after cracking decreases more rapidly for the beams with synthetic fibres due to these fibres low modulus of elasticity. 7

8 a) b) c) Figure 5. Short-time load deflection relations for the beams in Series I (a)steel fibre reinforced beams, (v f =0.5%), (b)synthetic fibre reinforced beams (v f =1.0%). (c)synthetic fibre reinforced beams (v f =0.7%) 8

9 Load (kn) Load (kn) BEFIB2012: Terje Kanstad, Giedrius Žirgulis a) b) CMOD (mm) Figure 6. Pre-cracking results from Series I. (a) Synthetic fibre reinforced beams (v f =1.0%), (b) Steel fibre reinforced beams (v f =0.5%) 5.3 Time dependent deflections The measured deflections in both series are presented in Figure 7. While the considered time period for the first series is 5.1 year, it is only about 3 months for the second. The latter is however still running, and will continue for at least one year. From the curves it can be concluded that the time shapes are similar for all the investigated fibre types. Furthermore it is seen that there is a considerable difference between the two series I curves for the beams with 1.0% synthetic fibres, and it is relevant to relate this to different fibre content in the two beams: 0.84% and 0.68% as shown previously in Table 2. This means that the ratio between fibre areas is 1.24, while the corresponding inverse ratio between the total deflections is about 1.4, and consequently the real fibre area can explain the deviations between the two specimens. It is also seen that the two beams with 0.5% steel fibres in Series II confirm the results for the corresponding beam in the first series. It should be noted that the load for the two SFR beams in Series II is about 7% lower than in Series I due to slightly lower residual strength. The basalt fibre reinforced beams have considerably lower deflections due to lower residual strength, and correspondingly lower long-time loads. As mentioned previously this is mainly due to this fibres low aspect ratio. In Figure 8, the calculated and measured creep ratios are compared. The methods are previously described in Chapter 4. It is seen that the agreement is surprisingly good, and related to calculation methods in general, it means that creep take place both in the compressive and in the tensile zone of the beams. Figure 9 presents the deflections of the four beams in Series I in an alternative way - as loaddeflection relations, and introducing the fact that for this beam geometry the CMOD and the deflections are about equal, it is seen that the total deformations are relatively small compared to the deflections and the crack-widths in the test used to determine the post-cracking strength as shown in Figure 5. This confirms that the conditions of the present test program are reasonable because they 9

10 are related to the service limit states, where the crack widths and deflections should be limited. a) b) Figure 7. Time dependent deflection development for the longtime test. (a) Series I, (b) Series II. 10

11 Figure 8. Measured and calculated development of the creep coefficients for the beams in Series I. Figure 9. Load deflection relations for the long time test series I. 6 DISCUSSION AND CONCLUSIONS A five year study of pre-cracked synthetic and steel fibre reinforced concrete beams has been carried out, and the results are being verified by a second test series which presently is running. From the results it can be concluded that the time development of the beam deflections is similar for all the investigated fibre types. After five years under load, the time dependent deflections are in the range 2-3 times the initial deflections, and these results can be closely related to creep coefficient values calculated according to the creep model in Eurocode 2. Evaluation of the experimental results based on a nonlinear hinge model is presently ongoing, and if this model shall describe the time dependent deflections reasonably well, creep in the tensile zone has to be included 11

12 The long time load level is 50% of the residual strength at 1.5mm CMOD, which is considerably higher than what usually will be the case in real concrete structures. Considering the test conditions further, it is important to note, that the failure mechanism of the considered fibre types in ordinary strength concrete, is fibre pullout, similarly as for most types of steel fibres. It is important to be aware of that the results might be quite different for fibres which are relatively stronger anchored in the concrete. Mapping of the real fibre content and the real fibre area crossing the critical cross section is an essential part of the result evaluation, and has been used to explain the different behaviour of similar specimens. Comparing residual strength test results of beams before and after the long-time tests, it can be concluded that the long-time testing, as carried out in the present investigation, does not have any significant influence on the residual strength. The major consequence of the present results is that creep due to long time load should be accounted for in calculations of crack widths and deformations in the service limit states. ACKNOWLEDGEMENTS The paper is based on the work performed in COIN - Concrete Innovation Centre ( - which is a Centre for Research based Innovation, initiated by the Research Council of Norway (RCN) in The Centre is directed by SINTEF, with NTNU as a research partners and with the present industrial partners: Aker Solutions, Norcem, Norwegian Public Roads Administration, Rescon Mapei, Skanska, Spenncon, Unicon, Veidekke and Weber Saint Gobain. REFERENCES [1] Gjestemoen. A., and T. Arneberget, A.T.: Synthetic fibre reinforced concrete. Master thesis Department of structural engineering, NTNU, Trondheim, (in Norwegian) [2] EN14651:2005+A1:2007, Test method for metallic fibre conrete. Measuring the flexural tensile strength (limit of proportionality (LOP) residual). [3] EN :2004, Eurocode 2: design of concrete structures. [4] Sandbakk, S., Fibre Reinforced Concrete. Evaluation of test methods and material development. PhD-thesis. Department of structural engineering, NTNU, Trondheim, ISBN [5] Lambrechts, A.N.: Concrete for a new world. The technical performance of steel and polymer based fibre concrete. N.V. Bekaert, Belgium [6] MacKay, J. And Trottier, J.F.: Post-crack behaviour of steel and synthetic FRC under flexural loading [7] Bernard, E.S.: Creep of cracked fibre reinforced shotcrete panels [8] Elliot, K.S.: BarChip fibre creep tests laboratory test report. The University of Nottingham, School of civil engineering, Nottingham England