Paper No.

TRB #01-3486 DEVELOPMENT OF TESTING PROCEDURES TO DETERMINE THE WATERCEMENT RATIO OF HARDENED PORTLAND CEMENT CONCRETE USING SEMI-AUTOMATED IMAGE ANALYSIS TECHNIQUES

by

Moon C. Won, Ph.D., P.E. Transportation Engineer Supervisor Materials Section, Construction Division Texas Department of Transportation 125 E. 11th Street Austin, Texas 78701-2483 ph) 512-465-7502 fax) 512-465-3093 e-mail: [email protected] and Edward Morgan Geologist Materials Section, Construction Division Texas Department of Transportation 125 E. 11th Street Austin, Texas 78701-2483 ph) 512-465-7350 fax) 512-467-3897 e-mail: [email protected]

A paper submitted to the 2001 Annual Meeting of the Transportation Research Board

ABSTRACT This paper presents the development of a semi-automated test procedure to determine watercement ratio of hardened concrete. The principle behind this test procedure is that water-cement ratio determines the porosity in concrete and porosity can be quantified by the intensity of the fluorescence light. Epoxy with special dye is used to impregnate the concrete. A thin section of concrete is made after the epoxy impregnation.. An image is captured and fluorescent intensity is measured. This fluorescent intensity is compared with a calibration curve to determine the watercement ratio. In the laboratory study, the effect of several test variables was evaluated. Fluorescent intensity decreases rapidly with exposure to ultraviolet (UV) light. The amount of dye and thickness of the thin section have an effect on fluorescent intensity. Based on this finding, a standardized test procedure was developed. Evaluation of this test method using mortar with various water-cement ratios indicates that fluorescent intensity is sensitive enough to water-cement ratios that water-cement ratios of hardened concrete can be accurately evaluated. This test method has been implemented for primarily forensic investigations with satisfactory results. Key Words: water-cement ratio, porosity, hardness, PCC, portland concrete cement, fluorescence, test methods

INTRODUCTION When concrete strength does not meet the minimum values required in the specifications, engineers in charge of the project have to make decisions as to what needs to be done. Often, those decisions are difficult to make unless the causes of the low strength are known. The Materials Section of the Texas Department of Transportation (TxDOT) receives numerous requests for water-cement ratio determination of hardened PCC concrete since it is one of the critical parameters controlling concrete strength and durability. Our present method of determining water-cement ratio is a qualitative estimate that is based on the observation of microtextural properties of the cement paste, and therefore, is not adequate for the accurate determination of water/cement ratio. An accurate and reliable quantitative method needs to be developed to provide accurate water-cement ratio information to project engineers so that they can make reasonable decisions. Recognizing this need, TxDOT initiated a research project in 1997 to develop a test procedure for determining the water-cement ratio of hardened concrete. As an initial step of the research, the other 49 states were surveyed to evaluate their current practices concerning the determination of water-cement ratios in hardened concrete. Table 1 summarizes the survey results, which shows most of the states do not have established test procedures. Recent advances in microscopic analysis techniques have resulted in the technology or tools that can be used to evaluate characteristics of cement paste. Previously, the evaluation of certain cement paste characteristics such as variations in capillary porosity and micro-crack propagation patterns has been limited with other microscopic techniques. Petrographers are now using a technique referred to as epiflourescent microscopy. Epiflourescent microscopy evaluates a thin section of a sample that has been impregnated with an ultraviolet (UV) sensitive epoxy dye. The UV sensitive dye becomes excited when UV light is transmitted through the sample. This technique was developed to enhance or bring out properties or characteristics in the concrete, which are obscured or less visible under conventional microscopy viewing. The excited dye exhibits a bright yellowish green glow. Features such as microcracks, air voids, and the capillary porosity of the paste become highly visible when viewed with the epiflourescent microscope. Any voids, gaps, cavities or porous medium impregnated with this epoxy will exhibit fluorescence. The principle behind this research is to relate the intensity of the capillary porosity fluorescence to the water/cement ratio of the hardened concrete. This correlation is possible since the capillary porosity of the cement paste is directly related to the water/cement ratio. Image analysis software is used to measure the intensity of the fluorescent and to assign a numerical value to this intensity. The numerical value is then correlated with known water-cement ratios. This technique was proposed a few years ago by Danish researchers (1). Technical details of this method are well documented in a couple of papers (1, 2). What remained to be done was to develop a correlation between the fluorescence intensity and water-cement ratio using state-ofthe-art image analysis software. The semi-automated testing procedure using the image analysis program makes the test procedure more reliable, faster, and easier to use. STUDY OBJECTIVES The primary objective of this study was to develop a semi-automated test procedure for watercement ratio determination. The procedure needed to utilize the most recent image analysis technique and to automate most of the procedures in order to minimize operator variability and to speed up the testing.

Table 1 Results of State Surveys on Water-Cement Ratio Evaluation Practices State Alabama Part I*
Yes No

Part II**
Yes No

Remarks We intend to have an in-house petrographic laboratory to determine among other concrete parameter the water/cement ratio in HPCC.

Arizona Arkansas California

The water/cement ratio is calculated from the measured water; i.e., free water (absorption ASTM C642) and chemically combined water (loss @ 520C for 4 hr) after correcting for the aggregate absorption and the calculated cement content (ASTM C1084 or California Test 403 by atomic absorption) cement per the attached. Petrographic analysis to visually count air, paste, aggregate per ASTM C-457-90.

Colorado Florida Georgia Hawaii Idaho Illinois Iowa Kansas Kentucky Louisiana Maine

We have just recently acquired an automated air void analysis system that has the option of determining the w/c ratio. We currently track w/c ratios during batching of concrete mixtures by the documentation method. Absorption of aggregate, free moisture in sand and stone, free water in admixtures, water added during batching and at the job site. We enforce maximum w/c ratio. (.40 w/ silica fume, .42 Class A reg. conc., .49 Class S Footings/Underwater conc. Pt II - We have looked into the use of microwave oven test methods. We ran many tests using the microwave oven and got many different results. We determined that the time and expense of running the test was not worth the trouble. Similar testing was also done in New Hampshire DOT. W/c ratio is not part of our current QA-QC spec.

Maryland Massachuset ts Minnesota

Michigan

Pt I - We are in the process of requesting materials and equipment to get set up to determine w/c ratio using UV sensitive epoxy impregnated thin sections. Hopefully within a few months we can begin this process. Pt II - Our chemistry lab is working on developing a method based on ASTM C1084, the silica method, and loss on ignition. We have considered using a petrographic procedure involving the determination of aggregate, paste, and air voids by linear traverse measurement in conjunction with the determination of evaporable water by vacuum saturation, then applying a maturity factor to compensate for the age of the concrete. However, the maturity factor presently must be estimated from

Missouri Mississippi Montana Nebraska New Hampshire New Jersey New Mexico New York

construction records, which may or may not be available. [We are interested in this subject and any methods determined. Please give us a copy of any reports.]

Air void counts are only to measure entrained air in hardened concrete.

North Carolina Ohio Oklahoma Oregon Pennsylvania Rhode Island South Carolina Utah Virginia Washington West Virginia

With the weight of cement determined by the test method, and the known weights of sand and stone, the difference from the total weight is the weight of water. We have experimented with the Troxler w/c gauge for plastic concrete.

UV epoxy impregnated polished thick sections with epifluorescent microscope (reflected light) Note that we do not currently make w/c ratio determinations by any method on a regular basis and have no current research scheduled. When necessary, our petrographers do describe the characteristics of hardened concrete, i.e., textural characteristics, point counts, observation of air voids, etc., but normally not with the intent to determine the w/c ratio. We are not working on new methods. We have a research contract with the University of Wisconsin - Madison to evaluate existing w/c measurement technologies available, and determine the feasibility of using any of these as QC/QA field measurements on PCC paving projects.

Wisconsin

Wyoming

* Part I of the survey inquired about any current methods for determining water/cement ratios of hardened portland cement concrete. ** Part II of the survey requested information concerning research development relating to the quantitative measure of water/cement ratio of hardened portland cement concrete.

BRIEF DESCRIPTION OF THE EQUIPMENT NEEDED AND HOW IT WORKS Equipment Overall, the equipment needed for the testing is as follows: Fluorescence Episcopic Microscope Video-Imaging System Image Analysis Software The fluorescent episcopic microscope is equipped with a mercury bulb that emits a light source through a series of filters that directs the light into the thin section. The series of filters are necessary to provide a specific wavelength light source into the thin section. The first filter in line with the mercury bulb light source is the excitation filter. The next filter in line is the dichroic mirror filter that directs the light source downward into the thin section. When the filtered light enters the thin section, the UV sensitive dye in the paste capillary pores is excited and gives off longer wavelength light that is directed upward through the emission filter into the video auxiliary tube. The excited longer wavelength light is directed into the video camera and captured as an image in the imaging system where the green-tone values are measured. Procedure Generate green-tone image from fluorescent episcopic microscope. Capture image and import to image analysis software. Threshold out the aggregate, air voids, microcracks, and unhydrated cement particles. Measure the green-tone intensity. Import green-tone data to spreadsheet. 6. Repeat the above five steps for a thin section to include twenty random points in a grid type fashion. Figure 1 shows the captured image. Note that the air void is represented by bright green tone, while unhydrated cement particles have dark green tones and aggregates are represented by dark tones. Capturing this image and importing the image to the image analysis software was automated by developing macros. Efforts were made to automate the above Step 3 as well; however, it was quickly obvious that it was not possible. Automation was not possible because any set of criteria developed for RGB to separate paste capillary pores from aggregates or unhydrated cement was not applicable for different thin sections. 1. 2. 3. 4. 5.

FIGURE 1 Green-tone image of 0.5 water-cement ratio thin section. TESTING PROGRAM The first step in this investigation was to identify potential sources of variability and quantify their effects on the test results. This step was necessary to enhance the repeatability and accuracy of the test procedure. This investigation was conducted using mortar only. The variables investigated in this study so far are: (1) exposure time of thin section to UV light (2) the amount of dye added to the epoxy, and (3) thickness of the thin section. The test procedure was refined based on this investigation and was applied to the mortar and concrete with various water-cement ratios. PRESENTATION OF THE RESULTS Investigation of Variabilities Figure 2 illustrates the decaying of fluorescent intensity over time due to the exposure to UV light. The thin section was put on the microscope and UV light was turned on. Images were captured immediately after the light was turned on. Subsequent images were captured in one and two minutes, thereafter, every two minutes up to 20 minutes. After that, the light was turned off for two hours. The light was turned on again and images were captured immediately, with subsequent three images every minute. Figure 2 shows a rapid reduction in green-tone values within one minute regardless of water-cement ratio, after which the rate of decrease is not as rapid. A two-hour rest period resulted in the recovery of the green-tone values by some; however, the recovered value is far lower than the values within the first minute. In another experiment, the rest period was extended for three days to see whether the longer rest period would result in significant recovery of green-tone values. The three-day rest period resulted in only 7-point

increase in green-tone values, which is similar to the increase in the two-hour rest. This experiment shows that green-tone values of thin sections decrease due to the exposure to the UV light, and the decrease is semipermanent, which implies that the time of exposure to UV light should be standardized to minimize the test variability. Based on this finding, images for the rest of the study were captured after approximately 5 seconds of UV light exposure. For the production of thin sections, 0.5 mg of dye is added to 100 ml of epoxy. To evaluate the effect of the variation in dye amount on green-tone values, three levels of dye were added; 0.3 mg, 0.5 mg, and 0.7 mg to 100 ml of epoxy. Figure 3 shows the 20 green-tone values captured per thin section with various dye amounts. The water-cement ratio was 0.6. The lower dye amount results in significantly lower green-tone values. However, adding additional 0.2 mg of dye to the standard amount did not change the green-tone values significantly. This experiment illustrates that the amount of dye needs to be close to the standard amount. The standard thickness of the thin section is 30 micrometers. The effect of thin section thickness was evaluated with three thicknesses; 27, 30, and 35 micrometers. Figure 4 shows the 20 greentone values per thin section with various thicknesses. The water-cement ratio was 0.5. A ten percent decrease in thickness from the standard 30 micrometers results in about 30 % reduction in green-tone values. However, there is no substantial difference in green-tone values between standard and thicker sections. This experiment also illustrates the need for tight control over the thickness of the thin sections. As a result of this investigation, the following items were standardized: (1) images need to be captured at 5 seconds after exposure to UV light, (2) the amount of dye added to 100 ml of epoxy should be 0.5 mg +/- 0.05 mg, and (3) the thickness of thin sections shall be 30 +/- 1 micrometer. Based on this standardized test procedure, green-tone values were evaluated for mortars with three water-cement ratios. Type III cement and Grade 20-30 Ottawa sand were used and, as described earlier, 20 green-tone readings were made for each thin section. Figure 5 shows the green-tone values for each water-cement ratio. There appears to be quite an overlap in greentone values for three water-cement ratios. However, note that the total number of readings per each water-cement ratio is 20, and most of the readings are around the average value. Also note 2 that R value is over 0.85. Figure 6 shows the frequency distribution of the same green-tone values in Figure 5. This figure clearly shows minimum overlap of green-tone values for different water-cement ratios. This figure also shows the potential of this test procedure for accurately evaluating water-cement ratio in hardened concrete. Evaluation of Water-Cement Ratio for Mortar In order to evaluate the water-cement ratio of specific hardened concrete, control thin sections need to be prepared and a calibration curve developed using the same materials that were used for the concrete to be evaluated. The concrete making materials that may have an effect on green-tone values are (1) cement type, (2) sand, (3) fly ash, and (4) admixtures. It is necessary to know whether different calibration curves need to be developed when the type or source of the above four materials changes. In this phase of the research study, cement type and sand were evaluated using mortar. For the effect of sand type, Grade 20-30 Ottawa sand and locally available siliceous river sand were used. The absorption of Ottawa sand was determined at 0 % and therefore, it was assumed that Ottawa sand was at SSD condition. Figure 7 shows that the green-tone values for Ottawa sand are higher than those for locally available siliceous river sand are. At this point, it is not known why the sand type makes a difference. Note that each greentone point represents the average of 20 values from a thin section. The effect of cement type was evaluated using Type I and III cements for three water-cement ratios. There is no significant difference in chemical compositions between the two cement types used in this evaluation; the 2 major difference is the fineness. The Blaine fineness was 372 and 553 m /kg for Type I and Type III, respectively. Figure 8 shows Type I cement produces higher green-tone values than Type III cement. This experiment shows that control thin sections need to be prepared for different

cement and fine aggregate types. The effect of fly ash and admixtures will be evaluated in the near future. Evaluation of Water-Cement Ratio for Concrete Concrete using identical coarse and fine aggregates was made for three different water-cement ratios. Figure 9 illustrates the green-tone values of concrete for three water-cement ratios. As in mortar (see Figure 5), there is a strong correlation between water-cement ratio and green-tone values. This experiment illustrates a strong potential for this test method to estimate the watercement ratio of hardened concrete.

CONCLUSIONS Based on the results of this limited study, the authors have drawn the following conclusions: 1. Semi-automated test procedure refined in this study using fluorescent dye and UV light can estimate water-cement ratio of hardened concrete reasonably well. 2. Time of exposure to UV light, dye amount, and thin section thickness need to be tightly controlled. 3. Cement and fine aggregate type have significant effect on the green-tone values. 4. Even though there is an overlap in green-tone values between different water-cement ratios, there is a significant difference in average green-tone values.

REFERENCES 1. Ulla, H.J., Johansen, V. and Thaulow, N. Estimating the Capillary Porosity of Cement Paste by Fluorescence Microscopy and Image Analysis. Materials Research Society Symposium Proceedings. Vol 370. 1995, pp. 227-236 2. Elsen, J. Lens, N., Aarre, T, Quenard, D., and Smolej, V. Determination of the Water/Cement Ratio of Hardened Cement Paste and Concrete Samples on Thin Sections Using Automated Image Analysis Techniques. Cement and Concrete Research. Vol. 25, No. 5, 1995, pp. 827834

60

50

Green Tone Values

40

30 w/c = 0.6 20 w/c = 0.4 2-hour rest

10

0 0 2 4 6 8 10 12 14 16 18 20

Time of Exposure (minutes)

FIGURE 2 Decaying of fluorescent intensity over time due to exposure to UV light.

70 60 50 Green Tone Values 40 30 20 10 0 0.2

0.3

0.4

0.5 Dye Amount

0.6

0.7

0.8

Figure 3 Effect of dye amount on green tone values.

60

50

Green Tone Values

40

30

20

10

0 24 27 30 33 Thickness of Thin Section (micrometer) 36

FIGURE 4 Effect of thin section thickness on green tone values.

50 45 40 35 Green Tone Values 30 25 20 15 10 5 0 0.3 0.4 0.5 Water-Cement Ratio 0.6 0.7
y = 83.195x - 6.4922 2 R = 0.8565

Figure 5 Green tone values for various water-cement

10 9 8 7 Frequency (%) 6 5 4 3 2 1 0 22-25 25-28 28-31 31-34 34-37 37-40 40-43 43-46 46-49 Green Tone Range FIGURE 6 Frequency distribution of green tone for various water-cement
0.4 0.5 0.6

60 55 50 Green Tone Values Ottawa Sand 45 Silica Sand 40 35 30 25 20 0.3

0.4

0.5 Water Cement Ratio

0.6

0.7

FIGURE 7 Effect of sand type on green tone values.

60 55 50 Green Tone Values 45 40 Type III Cement 35 30 25 20 0.3 Type I Cement

0.4

0.5 Water Cement Ratio

0.6

0.7

FIGURE 8 Effect of cement type on green tone values.

80

70

y = 159.59x - 33.115 R = 0.8714

60

Green Tone Values

50

40

30

20

10

0 0.3 0.4 0.5 Water Cement Ratio 0.6 0.7

FIGURE 9 Effect of water cement ratio on green tone values for concrete.