Materials and Structures/Matériaux et Constructions, Vol. 31, January-February 1998, pp 27-35 nalysis of progressive damage to reinforced ordinary and high performance concrete in relation to loading. Konin, R. François and G. rliguie Laboratoire Matériaux et Durabilité des Constructions, INS-US, Toulouse, France SCIENTIFIC REORTS aper received: July 17, 1996; aper accepted: November 12, 1996 B S T R C T This paper deals with the effect of loading on the microstructure of reinforced high performance and normal concrete. In the case of reinforced concrete structures, cracking is the only visible sign of mechanical degradation and is accompanied by microscopic damage to the concrete. To account for this damage, experiments on three different concretes with compressive strengths (f c28 ) of 45 Ma, 80 Ma and 100 Ma were carried out. The loadless initial state of concrete and the damage state after loading were described using the single replica technique observed on SEM. The way in which the microcracks extended was revealed, with predominantly aggregate bond failure and crack branching occurring with cracks spreading out around the coarse aggregate under stable progressive fracture conditions. n increase in the specific area of microcracks was seen to result from an increase in load. This phenomenon could have an influence on the durability properties of reinforced concrete structures. R É S U M É Cet article traite de l influence du chargement mécanique sur la microstructure des bétons armés courants et à hautes performances. Dans le cas des ouvrages en béton armé, la fissuration qui est le seul signe visible de la dégradation mécanique est accompagnée par un endommagement microscopique du béton. our caractériser cet endommagement, un programme de recherche a été mené sur 3 bétons différents dont les résistances caractéristiques (f c28 ) sont de 45 Ma, 80 Ma et 100 Ma. L endommagement initial et l endommagement dû à la sollicitation mécanique sont caractérisés par la méthode de la simple réplique au MEB. Le développement de la microfissuration dans le cas où la propagation de la fissure est stable, correspond préférentiellement à des ruptures aux interfaces pâte-granulat et à une fissure s insinuant dans la pâte en pontant les granulats. Un accroissement de la densité de microfissuration a été mesuré en fonction de l accroissement du niveau de chargement. Cet endommagement pourrait avoir une influence sur la durabilité des ouvrages en béton armé. 1. INTRODUCTION Corrosion of rebars is the major cause of the deterioration of reinforced concrete structures. During the corrosion process, the increase in volume of rust products creates tensile stresses causing the secondary cracking and spalling of concrete. This can result in a reduced load-bearing capacity and thus considerably reduce the service life of concrete structures. mong the many factors affecting reinforced concrete s durability, chloride penetration remains one of the major causes of embedded steel corrosion. It is well-known that high strength concrete (HSC) offers better performance in terms of durability. However, in the case of reinforced concrete structures, the penetration of chlorides does not depend only on concrete transfer properties, but also on the loading applied, the state of strains, mainly characterized by the presence of cracking, and the exposure to an aggressive environment. Cracks, most of which are clearly visible, have readily been identified as playing a role in the development of this phenomenon. From this perspective, rule books tend to emphasize the significance of crack widths as a Editorial note rof. Ginette rliguie is a RILEM Senior Member and Chairlady of TC 163-TZ on Interfacial Transition Zone and roperties of Transfer and a member of TC 154-EMC on Electrochemical Techniques for measuring Metallic Corrosion in Concrete. 0025-5432/98 RILEM 27
Materials and Structures/Matériaux et Constructions, Vol. 31, January-February 1998 criterion for durability. These rules led to an increase in the quantity of steel reinforcement (50 kg/m 3 in 1950 to 160 kg/m3 in 1990) in concrete in order to control this cracking. The increased cost resulting from such measures has led to research being undertaken on the matter. The most recent developments with respect to ordinary reinforced concrete show that cracks are not the major factor in the corrosion process, as long as their width does not exceed 0.5 mm. The concrete cover quality and the cover width seem to play the most significant role. Nevertheless, in the case of ordinary concrete, there is a correlation between the quality of concrete and cracking, with cracking providing the only visible indication of concrete damage. reas of the concrete structure affected by cracking imply that there is damage and that the concrete is of poorer quality. In order to take into account these different parameters, experiments on reinforced concrete elements were performed over a long period. These investigations compared ordinary and high strength concretes in order to quantify the durability properties in the presence of cracks. The results presented in this paper deal with the study of both initial damage and damage after loading in the pre- or post-cracking stage for ordinary concrete (45 Ma), HSC (80 Ma) and VHSC (100 Ma). The final objective of this study will be to determine the relative importance of the different parameters in reinforced concrete durability and to establish the influence of the service cracking on rebar corrosion development. 2.2 Test methods Two set-ups were used. The first one involved 11 22 cm cylinders of concrete reinforced with a centered rebar protruding from one side of the sample (Fig. 1). The reinforcing steel was loaded in tension. Because of the pull-out of the steel, the loading led to the formation of longitudinal cracking. The initial state of the concrete was assessed using the replica technique [1]. The microscopic damage due to loading both before and after the appearance of the macroscopic crack was also assessed by the replica technique. In the following comments, this model is referred to as the longitudinal model. The second set-ups used prismatic samples of 10 10 50 cm also reinforced with a centered rebar which protruded from each side of the concrete (Fig. 2). The steel bar was subjected to a tensile stress on both sides. This loading led to the formation of one or many transverse cracks in relation to the loading level. The changing state of damage was also assessed using the replica technique. In the following comments, this model is referred to as the transverse model. 2. EXERIMENTL ROGRM 2.1 Reinforced concrete specimens ll batches of samples were made up using the same three concrete compositions. These concrete compositions (Table 1) were designed to obtain concretes of 45, 80 and 100 Ma as characteristic compressive stresses (f c28 ). The cement used was an OC CEMI 52,5R. The first composition was used to make control specimens (ordinary concrete), while the second composition produced a high strength concrete (HSC) and the third composition produced a very high strength concrete (VHSC). In the second and third compositions, part of the cement was replaced by silica fume. Fig. 1 Sample with longitudinal crack. Table 1 Concrete Composition Control group HSC VHSC Gravel (kg/m 3 ) 1220 1166 1214 Sand (kg/m 3 ) 820 727 650 Cement (kg/m 3 ) 400 405 540 Silica fume (kg/m 3 ) 45 50 Superplasticizer (kg/m 3 ) 13.5 - (3%) 23.6 - (4%) Water (kg/m 3 ) 200 - (0.50) 157.5 - (0.35) 147.5 - (0.25) Compressive stress (Ma) (f c28 ) 45 80 100 Fig. 2 Sample with transverse crack. The two models led to two different types of cracking, but probably also to two different types of microcracking as both stress fields in the concrete, due to the load applied to the reinforcement, were significantly different. In the case of the transverse model, the concrete surrounding the reinforcement was subjected to a tensile 28
Konin, François, rliguie stress field, whereas in the case of the longitudinal model, the stress field in the concrete was more complex and a large part of the concrete was not in a tension state. In this paper, only the results obtained with the transverse model are presented. 2.3 Conservation mode ll samples were kept for 10 days in the curing room prior to the loading test. Then, they were kept in a closed container, in a saline and humid atmosphere subjected to wet and dry cycles every two weeks. Salt fog (35 g/l of NaCl) was generated by means of 4 sprays located in each upper corner of the confined chamber. 2.4 Experimental method Macroscopic and microscopic damage were monitored on the outside wall of the samples. The macroscopic observations led to the elaboration of cracking maps. The single replica technique was used to assess the evolution of the concrete s microstructure. For each sample, three levels of loading corresponding to the unloaded state, the pre-cracking state and the post-cracking state were studied (Table 2). Table 2 Tensile stress applied Control group HSC VHSC* Unloaded state 0 0 0 Loaded state before cracking 30 kn 30 kn 30 kn Loaded state until cracking 52.5 kn 60 kn 46.5 kn * For the VHSC, the smaller concrete cover (about 30 mm for VHSC vs. 42 mm for HSC and ordinary concrete) led to a decrease in the breaking load and since the sawing was carried out just before the loading test, the initial state of VHSC corresponds to the case of the bulk concrete not being affected by selfdesiccation. 3. CONCRETE DMGE 3.1 Macroscopic damage - Cracking map 3.1.1 Initial state (prior to loading) We noticed that cracks were not visible on the surface of the samples. 3.1.2 Damage after loading The cracks due to the tensile stress applied to the reinforcement were basically located near the central part of the sample. The number of cracks appearing on the sample seems to depend on the strength of the concrete and on the concrete cover. The cracking map for each sample is shown in Fig. 3. The width of the visible cracks is between 0.05 mm and 0.1 mm. Fig. 3 Cracking maps of the specimens. 3.2 Microscopic damage The progression of the microcracking was studied by the replica technique and quantified by total projections [2]. The latter process can be summarized as follows. The SEM images were digitized and the microcracking thus extracted was mapped. Two essential parameters characterize the microcracking: firstly, the specific area quantifying the significance of the microcracking network, and secondly, the degree of orientation quantifying the microcracking anisotropy. For each concrete, progress in microcracking in relation to load was studied. For this purpose, an area liable to crack during loading was selected, and the sixty views taken thereof as the load was increasing enabled a correlation between load and cracking density to be established. 3.2.1 Test method Each replica (corresponding to an area of 3 cm 2 ) magnified by 200 was divided into 918 fields measuring 674 478 µm 2. For each concrete, we quantified an area of about 19.6 mm 2 (corresponding to 60 views per replica). Some previous work [3] provides justification for this choice to quantify the microcracking of concrete samples. This research work was conducted with both less significantly microcracked samples (S v < 0.5 mm -1 ) and more significantly microcracked samples (S v > 1.5 mm -1 ) (Fig. 4). s a result, by taking into account the time required to analyze each replica, the number of views used to analyze a replica was usually fixed at 60 statistical views (Fig. 5). The drawbacks of this method are as follows: Firstly, it is obvious that this method does not allow the evolution of the same area in relation to the loading to be studied because it is not possible to obtain the same position of the replica in the sample holder from one replica to another. 29
Materials and Structures/Matériaux et Constructions, Vol. 31, January-February 1998 Fig. 5 Map of 60 statistical views. Fig. 4 Specific area of microcracking in relation to the number of views. R R Fig. 6 aste () aggregate () or paste () rebar (R) interface for unloaded OC sample (polished until 5 µm as grain size). Is it damaged or not? Fig. 7 aste () aggregate () or paste rebar (R) interface for cracked OC sample (polished until 14 µm as grain size). Is it damaged? Total rea 19.22 mm2 7.49 mm 0.390/mm 0.248/mm Dens. m [=135] 0.203 Dens. M [=45] 0.294 Sv 0.496/mm µ. 22.28% Fig. 8 Microcracking map for OC sample before loading. Fig. 9 Microcracking map and analysis for HSC sample before loading. 30
Konin, François, rliguie Fig. 10-a Microcrack in the paste () (view no. 52). Fig. 10-b Microcrack at the paste () aggregate () interface (view no. 42). Fig. 11 Microcracking map for VHSC sample before loading. Fig. 12-a aste () aggregate () interface for VHSC in the unloaded case (view no. 32); interface is not damaged. Fig. 12-b aste () aggregate () interface for VHSC in the unloaded case (view no. 50); interface is not damaged. Secondly, the coordinates were predetermined, so the value of the specific area of microcracking depends on the position of the replica. Thirdly, assessing microcracking is a subjective matter dependent on the user; it is thus necessary to have a reference (a view of the initial state) to establish the difference between the states of microcracking. For these reasons, we proceeded to carry out an analysis of 60 views side by side allowing for the following: changes in the same area to be studied in relation to loading, regardless of the position, as there were no predetermined coordinates, problems in extracting microcracking at the pasteaggregate interface to be avoided; Figs. 6-7 show the Total rea 19.24 mm2 6.84 mm 0.356/mm 0.226/mm Dens. m [=45] 0.202 Dens. M [=175] 0.238 Sv 0.453/mm µ. 10.05% Fig. 13 Microcracking map and analysis for OC sample in pre-cracking stage. aste Fig. 14-a View no. 4, no microcrack observed. Fig. 14-b View no. 8, microcrack at the interface (-aggregate, -paste). Fig. 14-c View no. 12, microcrack in the paste (aste). 31
Materials and Structures/Matériaux et Constructions, Vol. 31, January-February 1998 Total rea 19.34 mm2 8.17 mm 0.422/mm 0.269/mm Dens. m [=135] 0.240 Dens. M [=60] 0.301 Sv 0.538/mm µ. 13.98% Fig. 15 Microcracking map and analysis for HSC sample in pre-cracking stage. Fig. 16-a View no. 6, microcrack at the paste-aggregate interface. Fig. 16-b View no. 12, microcrack at the paste ()-aggregate () interface. significance of the degree of polishing on the characterization of the paste-aggregate interface. This is not a statistical analysis but a local study of the microcracking in the concrete. 3.2.2 Initial state (prior to loading) For ordinary concrete (OC): The results are the same as those reported in previous work [4, 5] and enable us to conclude that there is no initial microcracking in the paste or at the paste-aggregate interface (at 0.1 µm resolution) in the reinforced concrete samples (Fig. 8). For HSC: The results reveal the presence of microcracks due to self-desiccation. It can be observed that the Total rea 19.37 mm2 11.52 mm 0.595/mm 0.379/mm Dens. m [=120] 0.355 Dens. M [=45] 0.412 Sv 0.757/mm µ. 9.31% Fig. 17 Microcracking map and analysis for VHSC sample in pre-cracking stage. Fig. 18-a aste-aggregate interface in the pre-cracking stage (view no. 32); interface is damaged (-aggregate, -paste). Fig. 18-b aste-aggregate interface in the pre-cracking stage (view no. 50); interface is damaged (-aggregate, -paste). paste-aggregate interface is damaged (Fig. 10). The specific area of microcracking is about 0.50 mm -1 (Fig. 9). For VHSC: The results [6] reveal the absence of microcracks in the paste (Fig. 11) due to self-desiccation. Other tests showed that self-desiccation leads to the formation of microcracks whose specific area is about 0.3 mm -1. This does not contradict our results since this previous density means that microcracks are found only in 6 or 7 views out of 60 statistical views. For this specific work, the set of 60 views are all adjacent. Therefore, it is possible that the observed zone does not present any microcrack (Figs. 12-a and 12-b). 32
Konin, François, rliguie Total rea 19.21 mm2 17.55 mm 0.913/mm 0.582/mm Dens. m [=90] 0.445 Dens. M [=175] 0.722 Sv 1.163/mm µ. 28.41% Fig. 19 Microcracking map and analysis for OC sample in the post-cracking stage. aste C C aste Fig. 20-a Crack and microcrack in the paste (view no. 12). Fig. 20-b Crack at the paste ()-aggregate () interface (view no. 14). Fig. 20-c Microcrack in the paste (view no. 18). Total rea 19.33 mm2 16.53 mm 0.855/mm 0.544/mm Dens. m [=5] 0.441 Dens. M [=90] 0.632 Sv 1.089/mm µ. 21.61% Fig. 21 Microcracking map and analysis for HSC sample in the post-cracking stage. Fig. 22-a Increase of the damage at the paste-aggregate interface (view no. 8). Fig. 22-b Iincrease of the damage at the paste-aggregate interface (view no. 42). 3.2.3 Damage after loading in the pre-cracking stage (prior to aggressive environmental action) For OC: The results reveal the presence of microcracks at the paste-aggregate interface (Fig. 14). Increasing the load leads to an increase in the density of microcracking. The specific area is about 0.45 mm -1 (Fig. 13). For HSC: The results reveal that the tensile stress applied to the reinforced samples leads to a slight increase in the density of microcracks (Fig. 16). The specific area of microcracking is about 0.54 mm -1 (Fig. 15). For VHSC: The results reveal that the tensile stress applied to the reinforced samples leads to an increase in 33
Materials and Structures/Matériaux et Constructions, Vol. 31, January-February 1998 Total rea 19.37 mm2 29.77 mm 1.537/mm 0.979/mm Dens. m [=80] 0.809 Dens. M [=170] 1.120 Sv 1.957/mm µ. 19.70% Fig. 23 Microcracking map and analysis for VHSC sample in the post-cracking stage. C Fig. 24-a Crack and microcrack in the paste (view n 32), (-aggregate, C-crack, -paste). Fig. 24-b Increase of the damage at the interface and along the aggregate (view no. 50) (-aggregate, C-crack, -paste). the density of microcracks which are basically located at the paste-aggregate interface (Fig. 18). The specific area of microcracking is about 0.76 mm -1 (Fig. 17). 3.2.4 Damage after loading in the post-cracking stage (prior to aggressive environmental action) For OC: The tensile stress applied to the reinforced steel leads to the formation of a macrocrack going from one paste-aggregate interface to another (Fig. 20) and to an increase in the density of microcracks which are both located in the paste and at the paste-aggregate interface (Fig. 19). The specific area of microcracking is about 1.16 mm -1. For HSC: The tensile stress applied leads to the formation of a macrocrack located in the paste and to an increase in the density of microcracks going from one paste-aggregate interface to another (Fig. 22). The specific area of microcracking is about 1.09 mm -1 (Fig. 21). For VHSC: The tensile stress applied to the reinforcement leads to the formation of a macrocrack going from one interface to another (Fig. 24-a) and to an increase in the density of microcracks which are basically located at the paste-aggregate interface(fig. 24-b). The specific area is about 2.0 mm -1 (Fig. 23). C 4. DISCUSSION s evidenced in some previous work, no desiccation or self-desiccation microcracks are visible on the ordinary concrete, contrary to the HSC in which microcracking is probably due to the formation of the superficial hydrous gradients during the drying of the samples (the replicas were made on polished and dry surfaces). The type of fracture [7] (in compressive or tensile stress) of HSC and VHSC is clearly revealed by the significant microcracking observed before the fracture of these samples (Fig. 25). It is often claimed that HSC is a more homogeneous material than OC and hence prone to a more brittle behavior. This is supported by the fact that the mechanical properties of aggregates and the mortar mix are less at variance than in the case of OC (however, in the case of this research program, the compressive strength of coarse aggregates is about 177 Ma). The difference, as compared with mortar, is significant (after a curing of 10 days). Hence, we conclude that the coarse aggregates still act as rigid inclusions in HSC and VHSC. Thus, during application of the load, microcracking for HSC and VHSC (Fig. 24) is mainly located at the paste- Fig. 25 Variation of the specific area of microcracking for each concrete with the loading. 34
Konin, François, rliguie aggregate interface, similarly to OC, even if it is generally assumed [8] that the use of silica fume to produce these concretes leads to a resorption of the interfacial transition zone (ITZ). Nevertheless, the macrocrack is partially located in the paste. 5. CONCLUSION This study confirms that self-desiccation and desiccation microcracking are absent from initial ordinary concrete and that the tensile loading applied on reinforced bars leads to damage to the paste as well as to the ITZ. In the case of HSC and VHSC, the damage can mainly be seen to be located at the paste-aggregate interface. The same results were reported by Taerwe [9] for some HSCs subjected to stable compressive loading. Other results were reported by Masse [10] in a study attempting to characterize the link between the fatigue of HSC and transport properties. The fatigue test leads to damage mainly located at the paste-aggregate interface. For HSC and VHSC, an increase in loading leads to a significant increase in the density of microcracking. This damage could lead to increased penetration of aggressive ions, such as chloride ions, as already shown in the case of ordinary concrete [11]. research program concerning this last point is in progress at LMDC. The first result of chloride penetration [12] obtained on HSC (80 Ma) shows that there is an increase in chloride penetration linked to the microstructural effect due to mechanical loading. This increase is small because of the absence of ITZ in the case of HSC, as opposed to the case of ordinary concrete where the presence of the ITZ, along with the increase in microcracking with the load, leads to a strong loading impact on the chloride penetration. REFERENCES [1] Ollivier, J.., non destructive procedure to observe the microcracks of concrete by scanning electron microscope, C.C.R. 15 (6) (1985) 1055-1060. [2] Ringot, E., utomatic quantification of microcracks network by stereological method of total projections in mortar and concrete, C.C.R 18 (1) (1988) 35-43. [3] Yssorche, M.., Microfissuration et durabilité des Bétons à Hautes erformances, Thèse de doctorat INS Toulouse (1995) 63-65. [4] François, R. and Maso, J.C., Microfissuration initiale d un béton de structure, nnales de l ITBT 457 (1987). [5] François, R. and rliguie, G., Reinforced concrete: Correlation between cracking and corrosion, in Second CNMET/CI International Conference on Durability of Concrete, S 126 - vol. II., Montreal 08/1991. [6] François, R., Konin,. and rliguie, G., Evolution of the microstructure of reinforced HC in relation to the loading and the influence on durability, in 4th International Symposium on utilization of High-strength / High-performance concrete, roceedings of an international conference, May 1996 (aris) no. 151. [7] Wecharatama, M. and Chimamphant, S., Bond strength of deformed bars and steel fibers in HSC, roceedings of the MRS Symposium, Boston, 1987. [8] Gjo ry, O., Monteiro, E. and Mehta,. K., Effect of condensed silica fume on the steel-concrete bond, CI Materials Journal (Nov-Dec. 1990) 573-580. [9] Taerwe, L. R., Bond fracture and crack propagation in high strength concrete, roceedings of Third International Symposium on Utilization of High-strength Concrete, Lillehammer, 20-24 June 1993. [10] Masse,., DE US, 1995. [11] François, R. and Maso, J.C., Effect of damage in reinforced concrete on carbonation or chloride penetration, C.C.R. 18 (6) (1988) 961-971. [12] Konin,., François, R. and rliguie, G., Influence of the service load on durability of reinforced concrete in the presence of chlorides, MRS Fall Meeting, 27/11-01/12/1995. 35