COLOR AND NUMERICAL IMAGING FOR AIR CONTENT OF

HARDENED CONCRETE

 Charles E. Buchanan Jr.

President ROAN Laboratories

 

Charles E. Buchanan III

Lab Manager, ROAN Laboratories

  

ABSTRACT

For many years ROAN has used a stereo microscope to determine the air content of hardened concrete by the linear traverse method. This procedure is time consuming, and only minimal information is obtained. ROAN purchased a monocular microscope, CCD camera, high resolution monitor, and a pentium computer system, along with Optimet software which enables us to make a large number of pictures, obtain resolution down to approximately 10 micrometers, determine percent air voids, and to make a histogram distribution of the air void system. This is all done with a color imaging system which is superior to a gray scale because it can differentiate between two different colors that might have the same gray scale. In addition, a macro is used to give a point count so that a relationship can be maintained with previous ASTM methods.

 

HISTORY

Air content has been of vital concern in concrete and mortar technology for sixty plus years. It was first discovered in the Hudson Valley of New York quite by accident. Some say that fish oil contaminated Portland cement, while others say that sterates, used to waterproof natural and masonry cements, was used to clean out the finish mill system when the balls became coated. An astute observer noted that the resultant concrete showed a marked improvement in freeze-thaw durability.

It was later found that vinsol resin was an excellent air entraining agent as far as durability was concerned. However, difficulties were encountered when different cements were manufactured to air entraining specifications. This was especially important in New York state which specified a given rate of addition rather than measuring the resultant air content. It was soon determined that if the vinsol resin was saponofied with sodium hydroxide, rather than this reaction being dependent upon the alkalis of the cement, a much more consistent product could be made.

Not long after the air entraining method using 20-30 mesh standard sand was developed in ASTM Committee C-1,(Currently C-85) researchers started looking at methods of determining the air content of the hardened concrete and mortar. Two methods evolved in Committee C-9, both covered in C-457, Microscopical Determination of Parameters of the Air-Void System in Hardened Concrete. One, a point count method provides a quantitative analysis of the air content, the paste content, and the aggregate content of the hardened paste. This method involves setting a grid on the specimen to be run, and then determining what is under each point of the grid, either air, paste, or aggregate. One problem with this method is that when the grid ends up on an air bubble, you don't know if it is a large bubble or small. If the grid is set small enough, then a large bubble will be counted at two or more points. However, the smaller the grid is set, the more tedious the method becomes due to the number of points going up by the square of the grid dimensions.

The other method, known as the linear traverse, uses a series of imaginary lines drawn across the surface of the hardened concrete, with a counter being run for the entire traverse. A second counter is started at the beginning and stopped at the end of each air bubble encountered, and a third started and stopped for paste. The advantage of the linear traverse is that it directly measures the chord length, rather than relying on the distance between the grids as in the point count.

However, there are some problems with this in that a linear traverse is a single line drawn through the concrete specimen. Except by chance, it does not measure the complete bubble size, but a chord, either to the top or bottom of the center, or to the left or right of a center line drawn through the diameter. However, you do arrive at a the number of air bubbles, as well as the average chord length which is a measured entity. The volume percentage of air is correct however as a two dimensional surface is equivalent to a three dimensional volume as far as percentages are concerned.

Powers developed the theory of the average spacing factor, which was an average of how close the air bubbles were to each other. A number was arrived at if the spacing factor was below this, adequate freeze thaw durability would be obtained.

This, however, is a laborious task, and requires many hours or peering into a stereo microscope, and requires considerable training to become accomplished in differentiation between an air void, a sand pull out, or a dark piece of aggregate which resembles a shadow.

To alleviate some of this tedium, we purchased a monocular scope, a CCD color camera, and imaging software to attempt to automate this task. The microscope, an adjustable focus model, with the camera mounted on top, and the light source mounted on the bottom, is shown in Figure 1 . The RBG output from the camera controller is wired into a Matrox MVP-AT frame grabber board which enables you to capture an image as a digital computer file.

Samples are prepared using a Buehler 12 inch diamond saw, and a Buehler 12 inch lap. For lapping, we use a diamond grid with diamonds of 68 and 35 micron grit, and a rough polish and medium polish wheel. The polish wheels were purchased from TRW. This equipment is shown in figures 2, and 3.

Once a sample has had sufficient polishing, it is then filled with white out in order to enable us to see the void within the concrete specimen.

A sample of concrete is shown in Figure 4 which has a field of view of 160 mm. You can note that there are air voids visible in this photograph. However, we would be able to only identify voids down to 1 60 micrometers, as this would be equivalent to one pixel. Figure 5 shows the same sample with a field of view of 90 mm. Figure 6 shows the same concrete with a field of view of 38 mm which means that one pixel would be equivalent to 38 micrometers.

Figures 7 through 11 show the sample at the magnification that we normally use, with a field of view of 5350 micrometers. This enables us to see a void of 5 micrometers at one pixel.

With the software that we have, we can identify a color in its RBG numerical equivalent, and then color each area of the picture that contains that color. With our color impregnation of the white-out, under normal lighting conditions, all voids have a color of between 175 to 256 red, 160 to 256 blue, and 180 to 256 green.

Once we have identified the air voids, we can then convert each void to a series of numbers a description of which is shown below: (See Table 1)

AREA - The scaled area of the void.

PERIMETER - The distance around the void.

X - POSITION - The X coordinate of the void.

Y - POSITION - The Y coordinate of the void.

CIRCULARITY - How close the void is to a true circle, which is 12.

LONGEST AXIS - The longest chord of the void.

WIDTH - Greatest width measured perpendiculary from the longest axis.

MEAN GRAY SCALE - The average gray scale of the void.

STANDARD DEVIATION GRAY SCALE - The gray scale variation.

VARIATION OF GRAY - The maximum minus the minimum gray.

In order to conserve time, and computer space, we normally only look at one fifth of the concrete specimen. This relates to approximately 100 pictures of a piece of concrete. The frame grabber board has such good resolution, that we are able to make the photographs on the fly, never stopping the linear traverse stage. In this concrete specimen which we are looking at, the 100 pictures represents one-fifth of the area, and shows around 13500 air voids. In order to compare the statistics, we divided the pictures in one half, showing approximately 7000 voids in each set, and arrived at 5.54% air in one, and 5.82% air in the other. The distribution curve for the specimens are also very similar, indicating that our results are representative of the concrete as a whole.

 

FIGURE 1

FIGURE 2

FIGURE 3

FIGURE 4

FIGURE 5

FIGURE 6

FIGURE 7

FIGURE 8

FIGURE 9

FIGURE 10

FIGURE 11

Click to enlarge figures 1 - 11

  

 

Table I Typical Spreadsheet Output from Imaging Analysis

Area Perimeter X-Position Y-Position Circularity Longest
Axis
Width Mean
Grey
Std.Dev.
Gray
Var. of Gray
0.03 0.76 25.6 21.2 16.4 0.18 0.29 232.1 9 4 87.9
0.01 0.69 21.9 21.1 38.4 0.10 0.29 233.0 8.4 71.2
0.19 1.67 19.4 20.9 14.4 0.52 0.54 243.9 11.9 142.1
0.03 0.69 7.4 21.1 17.2 0.19 0.24 237.6 8.1 65.3
2.17 5.90 13.0 19.3 16.0 1.62 1.85 246.5 9.5 90.3
0.66 4.79 8.7 19.4 34.7 1.05 1.37 241.2 12.3 151.3
0.11 1.38 14.5 19.7 17.9 0.37 0.44 240.8 12.1 147.1
0.02 0.63 22.9 19.5 23.1 0.17 0.25 230.9 10.1 101.0
1.60 6.96 19.9 18.0 30.4 1.23 2.49 248.3 9.4 88.6
0.51 2.91 6.0 18.8 16.8 0.77 0.95 242.6 9.7 93.2
0.48 4.53 4.6 18.4 42.4 1.03 1.17 241.3 12.4 154.1
0.09 1.26 18.1 18.5 17.4 0.38 0.42 238.3 11.3 127.6
0.03 0.80 25.3 18.5 21.2 0.18 0.30 228.7 7.6 58.2
0.21 1.76 14.0 18.3 14.9 0.52 0.58 244.8 10.9 118.1
0.09 1.12 24.2 18.4 14.0 0.36 0.37 242.2 10.8 117.2
0.10 1.63 18.4 18.0 26.6 0.43 0.46 233.6 11.9 142.1
0.06 0.92 14.8 17.7 14.6 0.28 0.32 236.9 11.1 122.5
0.02 0.69 9.2 17.6 20.1 0.11 0.28 231.7 12.1 146.7
0.04 0.78 15.1 17.5 14.3 0.21 0.28 235.9 11.2 125.6
0.16 1.53 18.3 17.3 14.5 0.46 0.49 236.2 11.5 131.7
0.02 0.61 12.1 17.3 17.3 0.15 0.24 231.1 7.7 59.1
0.06 1.01 25.7 17.2 16.4 0.27 0.33 232.2 8.2 67.7
0.19 2.43 15.7 16.8 31.2 0.34 0.94 238.8 10.9 117.7
0.06 0.95 18.7 17.0 15.4 0.27 0.33 236.5 9.3 85.9
0.05 1.01 19.3 16.6 21.2 0.23 0.38 228.1 8.6 74.5
0.06 1.01 26.2 16.6 15.8 0.32 0.33 240.8 11.1 123.2
0.16 1.56 5.8 16.3 14.8 0.51 0.49 245.2 8.8 77.0
0.09 1.31 5.3 15.7 19.8 0.35 0.46 238.5 11.7 136.2
020 1.82 9.5 15.3 16.4 0 45 0.63 244.0 9.8 96.4
0.05 0.95 24.0 15.4 17 3 0.26 0.33 238.1 12.2 147.6
0.09 1.35 25.2 14.8 19.6 0.34 0.44 233.9 11.0 122.0
0.48 3.76 3.9 14.5 29.3 0.79 1.01 234.6 13.9 192.9
0.04 0.84 22.7 14.7 16.1 0.23 0.28 236.5 10.8 116.3
0.07 1.49 26.5 14.3 33.1 0.29 0.51 235.6 11.8 139.5
0.02 0.61 16.0 14.4 15.5 0.15 0.24 237.3 10.3 106.3
0.12 1.36 22.1 14.3 15.0 0.40 0.47 240.0 9.8 96.2
0.05 0.89 26.1 14.1 15.5 0.23 0.33 237.0 12.9 165.6

ROAN Home