Vous êtes sur la page 1sur 8

Quantitative microcrack analysis of RCC suffered different freezing-thawing damage degrees by DIP technique

Shuguang LI (1), Wei FENG (1), Juntao TIAN (1), Gaixin CHEN (1) and Yihui LU (1)
(1) State Key Laboratory of Simulation and Regulation of Water Cycle in River Basin, China Institute of Water Resources and Hydropower Research (IWHR), Beijing, P.R. China

Abstract The deterioration process of roller-compacted concrete (RCC) which suffered freezing-thawing (FT) is a process of initiation and growth of microcracks. Its significant to establish the correlations between the macro mechanical properties or durability of RCC and the microcrack characteristics and the first step is to quantify the geometrical characteristics of microcracks such as length, width, area, etc. This paper presents a quantitative analysis of the microcrack characteristics in air-entrained RCC suffered different freezing-thawing damage degrees by digital-image-processing (DIP) technique. A vacuum fluorescent-epoxy impregnation method is employed to highlight microcracks in the RCC matrix. A MATLAB-based software package called QUANSMIC (QUantitative ANalysis System of MIcrocracks in Concrete) is developed to automatically extract and analyze the microcrack characteristics such as the length density, area density and crack width in slices of RCC specimens, whose FT damage degrees are around 0(suffered no FT cycles), 0.06, 0.15, 0.30 and 0.40 respectively. Results show that the length density, the area density, the maximum width and average width of microcracks all increase with the increase of FT damage degree. Exponential correlations are established between either the length density or the area density and the FT damage degree. The analysis also shows that as the FT damage degree increases the microcracks firstly grow in the outer areas and then gradually develop into the inner areas of the RCC specimens. Keywords: roller compacted concrete (RCC), quantitative microcrack analysis, freezing-thawing damage, digital image processing, vacuum epoxy impregnation

1.

INTRODUCTION

Roller compacted concrete (RCC) is widely used in dam structures for its lower material costs, higher construction speed and lower costs in post-cooling and formworks over conventional concrete. Owing to the relative stiff and dry consistency and lower cement content, RCC is different from conventional concrete in durability and mechanical/physical properties. When RCC dams are built in cold regions, its important and inevitable to predict the freezing-thawing (FT for short) durability of RCC. Researchers have done a lot of work on the effects of the mixing proportions such as the water/cement ratio, cement content, air content and air voids spacing factor on the FT resistance of RCC [1-4]. However, none of them focused on the initiation and growth process of microcracks, which is essential of the deterioration process of RCC subjected to FT. With the development of optical microscopy and digital-image-processing (DIP) techniques, some scientists such as Ammouche[5], Litorowicz[6], Soroushian and Elzafraney[7, 8] , Li and Chen[9] began to quantify microcrack patterns in conventional concrete or mortar deteriorated by mechanical loadings, frost and other factors to get better and deep insights into the damage and failure process of cementitious materials. But the development of microcrack characteristics in RCC during FT cycles hasnt been covered yet so far. This paper tries to quantitatively study the initiation and growth process of microcracks in RCC subjected to FT cycles by DIP technique. Firstly, a method for identification and quantification of microcrack patterns in concrete using fluorescent microscopy and image analysis techniques is presented. Then the microcrack characteristics including the length density, the area density, the maximum width and average width in RCC deteriorated by different FT cycles are obtained. Finally, the spatial distributions of cracks in the RCC specimens during FT cycles are also investigated. 2. FREEZING AND THAWING TESTS

The FT tests of RCC specimens and the resonant frequency tests of the specimens at different FT cycles were done according to the Test code for hydraulic concrete (SL352-2006) of China [10], which is similar with Standard Test Method for Resistance of Concrete to Rapid Freezing and Thawing(ASTM C 666/C 666 M-03(2008)). The mixing proportion of RCC was shown in Table 1. The air content in the RCC specimens is 3.6% and the compressive strength at 28d is 24.2MPa. Table 1 Mixing proportion of RCC specimens
Water/ cement 0.45 Sand ratio (%) 33 Admixture (%) Super-pla sticizer 0.6 Air-entra inment 0.05 Water (kg/m ) 84.0
3

Cement (kg/m ) 93.3


3

Pozzolan (kg/m ) 93.3


3

Sand (kg/m ) 733.7


3

Aggregate (kg/m ) 1493.5


3

VC (s) 4.1

FT damage degree is proposed to describe the deteriorating extent of FT cycles to RCC and it is calculated by the following formula, DFT=1-E/E0 (1), where DFT is the FT damage degree, E and E0 are the dynamic moduli of elasticity of the damaged and undamaged RCC respectively.
2

The relative dynamic modulus of elasticity and FT damage degree of the RCC specimens at different FT cycles are shown in Fig. 1. It can be seen from Fig. 1 that the relative dynamic modulus of elasticity increases with the increase of FT cycles while the FT damage degree increases with the increase of FT cycles. The FT damage degree reaches as high as 0.39 when that FT cycle is 400. Quantitative microcrack analysis was performed on the RCC specimens after being tested for resonant frequency at specified FT cycles.

Fig.1 Relative dynamic modulus of elasticity and FT damage degree of RCC at different FT cycles

3.

QUANTITATIVE MICROCRACK ANALYSIS

3.1 SPECIMEN PREPARATION The prerequisite for quantitative microcrack analysis is an appropriate preparation of specimens to highlight microcracks against the matrix in the microscopic images to facilitate the following image processing procedures. Fluorescent epoxy vacuum-impregnation method is used for it has little influence on the original microcrack patterns in concrete. The concrete specimens for microscopic analysis were prepared by fluorescent epoxy vacuum-impregnation method in the following procedures: Three slices of 15 mm to 20 mm thickness were cut from the middle and the end parts of the 10 cm 10 cm 40 cm prismatic specimen which had been tested for resonant frequency at different FT cycles. The surfaces of the slices were ground and polished using a grinder to obtain a smooth flat surface. The smooth surfaces were rinsed with water at low and high speeds to remove any leftover debris. The slices were dried in an oven at a temperature of 40~50 C for 24 h. The slices were placed in a plastic container and evacuated for at least 1 h at a constant pressure of 1 kPa in a vacuum chamber. The plastic container inside the vacuum chamber was then filled with the epoxy resin containing fluorescent dye while maintaining the vacuum (See Fig. 2 (a)). Finally, the slices were taken out from the vacuum chamber and from the plastic container, and then slightly ground to remove the excess epoxy on the surfaces once the epoxy was hardened.
3

3.2 QUANTITATIVE ANALYSIS SYSTEM OF MICROCRACKS IN CONCRETE Besides specimen preparation, reliable quantitative analysis of microcracks requires acquisition, processing, and analysis of a large number of microscopic images because of the complexity and heterogeneity of the mesoscale structures of concrete. The impregnated and polished concrete slices, namely the microscopic specimens, were observed under a fluorescent microscopy to search for microcracks and have the images stored in the computer for further processing and analyzing (see Fig. 2(b)). The specimens were covered by a steel mesh which had 81 grids (the size of each grid is 1 cm*1 cm) and was helpful to seek and locate the microcracks. The specimens were scanned manually grid by grid on the stage at a magnification of 40 which was sufficient to detect microcracks as thin as 2 m in width. Every grid was carefully observed to find all the microcracks inside and the total observed area for each slice took up as high as 81% of the slices total area.

(a)

(b)

Fig.2 The vacuum impregnation system (a) and the fluorescent microscope system (b)

A MATLAB-based software package called QUANSMIC (Quantitative Analysis System of Microcracks in Concrete) was developed to acquire the geometrical characteristics of microcrack patterns. It includes four modules, namely image-binarizing module, shape-analyzing module, crack-skeletonizing module and automatic-measuring module. The original colour microscopic images containing microcracks are firstly converted into greyscale images, and then the microcracks are recognized and extracted by the image-binarizing module. Noises and voids are distinguished from the microcracks and deleted by the shape-analyzing module and then the preserved microcracks are skeletonized by the crack-skeletonizing module. Finally the geometrical characteristics such as area, length and average width of the microcracks are measured and calculated by the automatic-measuring module. The precision of QUANSMIC was validated with several ordered specimens with shapes of known sizes and the comparisons showed that QUANSMIC has a high accuracy. Details about the QUANSMIC are introduced in ref [9]. The processing flowchart of two typical microcrack images by QUANSMIC is shown in Fig. 3.

(a)

(b)

(c)

(d)

Fig. 3 Processing flowchart of microcrack images by QUANSMIC (Agg: Aggregate, Mor: Mortar): a) grayscale images; b) binary images after thresholding; c) after shape analysis; d) skeleton of the microcracks

4.

QUANTITATIVE MICROCRACK ANALYSES OF FT-DAMAGED RCC

Six microcrack characteristic parameters, namely the total length, length density, total area, area density, maximum width and average width are calculated by QUANSMIC to quantify the microcrack patterns in RCC specimens subjected to different FT degrees. They are calculated as follows: Total length - sum of the microcrack lengths on all the microscopic images. Length density - ratio of the total length to the observing area. Total area - sum of the microcrack areas on all the microscopic images. Area density - percentage of the area of the microcracks to observing area. Average width - total area divided by the total length. Maximum width - the maximum value of all the average microcrack widths. 4.1 Evolutions of microcrack characteristics with FT damage degrees The evolutions of the length density, the area density, the average width and maximum width with FT damage degrees are shown in Fig. 4(a) ~(c). The temperature distributions along the axial directions of the concrete specimens kept in the freezing-thawing machine are uniform, so it is supposed that the microcrack patterns in three cross sections are representative of the microcrack patterns in the whole specimen. The variance of the microcrack characteristics in the three slices of a sample was small indicating that the analysis results were statistically reliable. It can be seen from Fig. 4(a) and (b) that the both the length density and the area density increase with the increase of the FT damage degree. A rapid and sudden increase in the microcrack density (length density/area density) could be found when the FT damage degree is over 0.25. In addition, exponential correlations between either the length density or area density and the FT damage degree are established and the correlation coefficients are both no less than 0.80. It can also be seen from Fig. 4(a) ~ (b) that for the RCC specimens with zero damage degree (without any deterioration), either the length density or the area density of microcracks is at a very low level. It indicates that a few initial microcracks could be found in the undamaged concrete
5

before any loading or deterioration because of the shrinkage of paste during the cement hydration process. It can be seen from Fig. 4 (c) that both the maximum width and the average width of microcracks increase with the increase of the FT damage degree. However, as the FT damage degree increases from 0 to 0.39, the maximum crack widths increases dramatically from around 10 m to 35 m, while the average width increases gradually from around 7 m to around 16 m. The reason is probably that while the existing microcracks become wider and wider as the FT damage increases, many new microcracks are emerging which makes the average width of all the microcracks changed a little.

(a)

(b)

(c) Fig. 4: Evolutions of microcrackcharacteristic parameters of concrete with different FT damage degrees: a) length density; b) area density; c) maximum width and average width.

4.2 Evolutions of the spatial distribution of cracks in RCC specimens Each microcrack image is saved into the computer and named after its coordinate in the positioning mesh which divides the RCC slice into 81 blocks (99) of 1 cm1 cm in size. The 81 blocks in each slice are then divided into two areas, namely the outer area and the inner area (see
6

Fig.5 (a)). Accordingly, the microcracks are divided into two parts, the outer cracks (microcracks lying in the outer area) and the inner cracks (microcracks lying in the inner area). Fig. 5(b) and (c) show the evolutions of the length percent (total length of outer cracks divided by total length of all the cracks) and the area percent of the outer cracks (total area of outer cracks divided by total area of all the cracks) to all the cracks on the slices respectively. It can be seen from Fig. 5(b) and (c) that for undamaged RCC specimens, either the length percent or area percent of the outer cracks is around 50%, indicating that the spatial distribution of initial microcracks in the cross slices of RCC specimens are uniform and random. With the increase of FT damage degree, either the length percent or the area percent of the outer cracks firstly increases from around 50% to 70% or even close to 80% and then gradually decreases to around 60%. It indicates that with the increase of FT damage degree, microcracks firstly emerge and develop in the outer area (marginal area) of the RCC specimens and then gradually grow into the inner area after the FT damage degree is over 0.25.

(a)

(b)

(c)

Fig. 5: Schematic of the division of the inner area and the outer area of each slice (a); evolution of the length percent of outer cracks with FT damage degree (b); evolution of the area percent of outer cracks with FT damage degree (c)

5.

CONCLUSIONS

Based on the DIP technique, a thorough investigation was made on the development of microcrack patterns in RCC specimens suffered five different FT damage degrees, 0, 0.06, 0.15, 0.30 and 0.40. The following conclusions can be drawn from the results and analysis above. (a) The length density, area density, maximum width and average width of microcracks all increase with the increase of FT damage degree in RCC.
7

(b) Both the length density and area density have exponential correlations with the FT damage degree. The corresponding correlation coefficients are no less than 0.80. (c) As the FT damage degree increases the microcracks firstly grow in the outer areas and then gradually develop into the inner areas of the RCC specimens. (d) When the FT damage degree in RCC increases to over 0.25, rapid and sudden increases in the length density and area density of microcracks could be found, and the length percent as well as the area percent of the outer cracks begins to decrease. It indicates that there may be a turning point during the FT damaging process in RCC when the microcracks begin to emerge and grow dramatically from the outer areas to the inner areas of RCC. It needs to be verified by more studies in the future.

ACKNOWLEDGEMENTS Financial support from the Ministry of Water Resources of P. R. China (Grant No. 200701041) is gratefully acknowledged. REFERENCES
[1] Pigeon M, Malhotra V M. Frost resistance of roller-compacted high-volume fly ash concrete. Journal of Materials in Civil Engineering, 1995, 7(4): 208-211. [2] Gao P-W, Wu S-X, Lin P-H, et al. The characteristics of air void and frost resistance of rcc with fly ash and expansive agent. Construction and Building Materials, 2006, 20(8): 586-590. [3] Hazaree C, Ceylan H, Wang K. Influences of mixture composition on properties and freeze-thaw resistance of rcc. Construction and Building Materials, 2011, 25(1): 313-319. [4] Hazaree C, Wang K, Ceylan H, et al. Capillary transport in rcc: Water-to-cement ratio, strength, and freeze-thaw resistance. Journal of Materials in Civil Engineering, 2011, 23(8): 1181-1191. [5] Ammouche A, Breysse D, Hornain H, et al. A new image analysis technique for the quantitative assessment of microcracks in cement-based materials. Cement and Concrete Research, 2000, 30(1): 25-35. [6] Litorowicz A. Identification and quantification of cracks in concrete by optical fluorescent microscopy. Cement and Concrete Research, 2006, 36(8): 1508-1515. [7] Soroushian P, Elzafraney M, Nossoni A. Specimen preparation and image processing and analysis techniques for automated quantification of concrete microcracks and voids. Cement and Concrete Research, 2003, 33(12): 1949-1962. [8] Elzafraney M, Soroushian P. Quantitative microstructural investigation of deteriorated reinforced concrete bridge deck. Journal of Materials in Civil Engineering, 2005, 17(2): 159-167. [9] Li S, Chen G, Lu Y. Quantitative evaluation of freezing-thawing damage in plain concrete by digital-image-processing technique. Proceedings of the 2nd International Conference on Microstructure Related Durability of Cementitious Composites, Amsterdam, the Netherlands, 2012. [10] China Institute of Water Resources and Hydropower Research. Test code for hydraulic concrete, sl352-2006, 2006.

Vous aimerez peut-être aussi