Development and Testing of a Capacitor Probe to Detect Deterioration in Portland Cement Concrete

Transcription

1 Development and Testing of a Capacitor Probe to Detect Deterioration in Portland Cement Concrete by Brian K. Diefenderfer Thesis submitted to the Faculty of the Virginia Polytechnic Institute and State University in partial fulfillment of the requirements for the degree of Master of Science In Civil Engineering APPROVED Dr. Imad L. Al-Qadi, Chair Professor of Civil Engineering Dr. Sedki M. Riad Dr. Gerardo W. Flintsch Professor of Electrical Engineering Assistant Professor of Civil Engineering September 1998 Copyright 1998, Brian K. Diefenderfer

2 Development and Testing of a Capacitor Probe to Detect Deterioration in Portland Cement Concrete Brian K. Diefenderfer Chair: Dr. Imad L. Al-Qadi The Via Department of Civil and Environmental Engineering Abstract Portland cement concrete (PCC) structures deteriorate with age and need to be maintained or replaced. Early detection of deterioration in PCC (e.g., alkali-silica reaction, freeze/thaw damage or chloride presence) can lead to significant reductions in maintenance costs. Portland cement concrete can be nondestructively evaluated by electrically characterizing its complex dielectric constant in a laboratory setting. A parallel-plate capacitor operating in the frequency range of 0.1 to 40.1 MHz was developed at Virginia Tech for this purpose. While useful in research, this approach is not practical for field implementation. In this study, a capacitor probe was designed and fabricated to determine the in-situ dielectric properties of PCC over a frequency range of 2.0 to 20.0 MHz. It is modeled after the parallel-plate capacitor in that it consists of two conducting plates with a known separation. The conducting plates are flexible, which allows them to conform to different geometric shapes. Prior to PCC testing, measurements were conducted to determine the validity of such a system by testing specimens possessing known dielectric properties (Teflon). Portland cement concrete specimens were cast (of sufficient size to prevent edge diffraction of the electromagnetic waves) having two different air contents, two void thicknesses, and two void depths (from the specimen s surface). Two specimens were cast for each parameter and their results were averaged. The dielectric properties over curing time were measured for all specimens, using the capacitor probe and the parallel-plate capacitor. The capacitor probe showed a decrease in dielectric constant with increasing curing time and/or air content. In addition to measuring dielectric properties accurately and monitoring the curing process, the capacitor probe was also found to detect the presence and relative depth of air voids, however, determining air void thickness was difficult. ii

3 ACKNOWLEDGEMENT The author expresses his gratitude to his advisor, Dr. Imad L. Al-Qadi for contributing to this opportunity to conduct research and providing the guidance necessary to undertake such a research project. In addition, thanks are given to the committee members, Drs. Sedki M. Riad and Gerardo W. Flintsch, for giving their time and expertise to help in the completion of this work. Thanks are given to Jason Yoho for his friendship, hard work, and patience as a research partner. His help in the author's overall understanding and assistance in specimen preparation were instrumental to the completion of this work. The author extends his heartfelt appreciation to his family for their unending support and understanding. Additionally, the assistance of the author's colleagues, Amara, Salman, Walid, Alex, Erin, James, Kiran, Ramzi, and Stacey from the Civil Engineering Materials Program and Iman from the Electrical Engineering Time Domain Lab greatly contributed to the author's knowledge as well as the progress of this research. The author also recognizes the support and interest received from the United States Air Force Kirtland Base contract # F29650-W0542 under the direction of John Rohrbaugh and Wes Tucker and the National Science Foundation grant # MSS under the direction of Ken P. Chong. iii

4 TABLE OF CONTENTS LIST OF FIGURES vii LIST OF TABLES ix CHAPTER 1 INTRODUCTION Background Problem Statement Objectives of Research Scope of Research 4 CHAPTER 2 BACKGROUND Portland Cement Concrete Hydration of Portland Cement Hydrated Cement Paste Aggregate Phase Transition Zone Deterioration of PCC Corrosion of Reinforcing Steel in PCC Alkali-Silica Reaction Freeze-Thaw Damage Dielectric Materials Polarization Concepts 25 CHAPTER 3 PARALLEL-PLATE MEASUREMENT SYSTEM System Design and Setup Theoretical Background of the Parallel-Plate Capacitor Parallel-Plate Capacitor Model 34 iv

5 3.4 Parallel-Plate Calibration Standards Open Calibration Standard Load Calibration Standard Short Calibration Standard Equations Governing the Parallel-Plate System Parallel-Plate Measurement System Calibration Determination of Remaining Unknowns Calibration Schemes 52 CHAPTER 4 CAPACITOR PROBE MEASUREMENT SYSTEM Physical Construction of the Capacitor Probe Capacitor Probe Design Capacitor Probe Plate Configurations Capacitor Probe Calibration Standards Open Calibration Standard Load Calibration Standard Short Calibration Standard Known Dielectric Material Calibration Standard Equations Governing the Capacitor Probe System Load Calibration Open Calibration Calibration Using Material of Known Dielectric Constant Short Calibration Determination of Remaining Unknowns Correction Function 75 CHAPTER 5 TESTING PROGRAM Specimen Preparation Dielectric Constant Measurements 80 v

6 CHAPTER 6 DATA PRESENTATION AND ANALYSIS Discussion of Data Parallel-Plate Capacitor vs. Capacitor Probe Final Remarks 94 CHAPTER 7 SUMMARY AND CONCLUSIONS Findings Conclusions 96 CHAPTER 8 RECOMMENDATIONS 97 REFERENCES 98 APPENDIX A 107 APPENDIX B 110 APPENDIX C 114 APPENDIX D 145 APPENDIX E 154 vi

7 LIST OF FIGURES Figure 2.1 Typical atom in (a) the absence of and (b) under an applied field (after Balanis, 1989) 22 Figure 2.2 Parallel-plate capacitor in the presence of (a) a vacuum and (b) dielectric material (after Callister, 1994) 23 Figure 2.3 Representation of dielectric polarization: (a) ionic (b) electronic, (c) dipole, (d) heterogeneous (after Jastrzebski, 1977) 26 Figure 2.4 A model representation the molecular interaction effect (after Debye, 1929) 28 Figure 3.1 Schematic setup for the parallel-plate capacitor 32 Figure 3.2 Electric field distribution between the two plates of the parallel-plate capacitor 33 Figure 3.3 Parallel-plate capacitor with and without specimen under test 35 Figure 3.4 Parallel-plate load calibration standard 39 Figure 3.5 Large height calibration standard base 39 Figure 3.6 Small height calibration standard base 40 Figure 3.7 Short calibration standard 40 Figure 3.8 Schematic of parallel-plate measurement system and model. 41 Figure 3.9 (a) General parallel-plate system model and (b) general S-parameter model 43 Figure 3.10 (a) Load parallel-plate system model and (b) load S-parameter model _ 44 Figure 3.11 (a) Open parallel-plate system model and (b) open S-parameter model 46 Figure 3.12 (a) Short parallel-plate system model and (b) short S-parameter model_ 47 Figure 3.13 (a) MUT parallel-plate system model and (b) MUT S-parameter model _ 50 Figure 3.14 Real part of dielectric constant of nylon using different calibrations 53 Figure 3.15 Imaginary part of dielectric constant of nylon using different calibrations 54 Figure 4.1 The capacitor probe 55 Figure 4.2 Schematic of EM field distribution at (a) high frequency and (b) low frequency 56 Figure 4.3 Capacitor probe load calibration standard 60 Figure 4.4 Capacitor probe short calibration standard 61 vii

8 Figure 4.5 (a) General capacitor probe model, (b) general S-parameter model of the interface network, and (c) general S-parameter model of the combined network 63 Figure 4.6 (a) Load capacitor probe model and (b) load S-parameter model 64 Figure 4.7 (a) Open capacitor probe model and (b) open S-parameter model 65 Figure 4.8 (a) Material capacitor probe model and (b) material S-parameter model 67 Figure 4.9 (a) Short capacitor probe model and (b) short S-parameter model 68 Figure 4.10 (a) MUT capacitor probe model and (b) MUT S-parameter 74 Figure 5.1 Schematic of Styrofoam placement in PCC slabs 80 Figure 5.2 Dielectric properties of PCC measured with capacitor probe comparing Styrofoam and an Air Bag used to apply a systematic and repeatable pressure 81 Figure 5.3 Dielectric properties of Teflon measured with parallel-plate capacitor 82 Figure 6.1 Average dielectric properties (real part) for type A specimens at 42 days after mixing 83 Figure 6.2 Average dielectric properties (imaginary part) for type A specimens at 42 days after mixing 84 viii

9 LIST OF TABLES Table 3.1 Reflection coefficients from the measured calibration standards 49 Table 4.1 Plate size and spacing of the different capacitor probes 59 Table 4.2 Reflection coefficients from the measured calibration standards 69 Table 5.1 Mix design and specimen characteristics 79 Table 6.1 Dielectric constant for type A specimens (2% air content, w/c = 0.45) measured using capacitor probe a 84 Table 6.2 Dielectric constant for type C specimens (6% air content, w/c = 0.45) measured using capacitor probe a 85 Table 6.3 Dielectric constant for type A specimens (2% air content, w/c = 0.45) measured using the parallel-plate capacitor 85 Table 6.4 Dielectric constant for type C specimens (6% air content, w/c = 0.45) measured using the parallel-plate capacitor 85 Table 6.5 Dielectric constant for type D specimens (6% air content, w/c = 0.45, and 7.5 mm thick void at 25 mm depth) measured using capacitor probe a 86 Table 6.6 Dielectric constant for type E specimens (6% air content, w/c = 0.45, and 7.5 mm thick void at 50 mm depth) measured using capacitor probe a 86 Table 6.7 Dielectric constant for type F specimens (6% air content, w/c = 0.45, and 15 mm thick void at 25 mm depth) measured using capacitor probe a 87 Table 6.8 Dielectric constant for type G specimens (6% air content, w/c = 0.45, and 15 mm thick void at 50 mm depth) measured using capacitor probe a 87 Table 6.9 Dielectric constant for type D specimens (6% air content, w/c = 0.45) measured using the parallel-plate capacitor 87 Table 6.10 Dielectric constant for type F specimens (6% air content, w/c = 0.45) measured using the parallel-plate capacitor 88 Table 6.11 Difference in dielectric constant due to air content as measured using capacitor probe a 90 ix

10 Table 6.12 Difference in dielectric constant due to void depth (7.5 mm thick void) as measured using capacitor probe a 90 Table 6.13 Difference in dielectric constant due to void depth (15 mm thick void) as measured using capacitor probe a 91 Table 6.14 Change in dielectric constant due to void thickness (25 mm void depth) as measured using capacitor probe a 91 Table 6.15 Change in dielectric constant due to void thickness (50 mm void depth) as measured using capacitor probe a 92 Table 6.16 Change in dielectric constant of type A specimens (6% air, 0.45 w/c) due to different capacitor probes 93 Table 6.17 Change in dielectric constant of type G specimens (6% air, 0.45 w/c, 15 mm thick void at 50 mm depth) due to different capacitor probes 93 x

11 CHAPTER 1 INTRODUCTION The ability of any industrialized nation to produce and sustain economic growth is directly related to its ability to transport the goods and services that it creates. Without a viable infrastructure system, basic public services (e.g., food distribution, water supply, waste removal, and medical facilities) cannot be effectively disbursed. The success of such a system depends on the ability of government policy makers to strike a balance between available funds and the need for repair or replacement of infrastructure components. This balance often proves difficult to achieve in a political climate that gauges success and progress with the creation of new facilities and not the rehabilitation of existing ones. In the United States there are more than 581,000 bridges in the national highway infrastructure system. Nearly 32% of these bridges are listed as structurally deficient or functionally obsolete (ASCE, 1998). Approximately forty-five percent of the bridges in the United States were constructed between the end of World War II and As these older structures approach (and exceed) their designed service life, a plan must be implemented to categorize the need for repair or rehabilitation, based on the level of deterioration associated with a given structure. 1.1 Background Portland cement concrete (PCC) is the most widely used construction material in the world due to its ease of preparation and molding, its low price (relative to other construction materials), and the abundance of its constituent materials (cement, aggregate, and water). However, there is more to PCC than simply mixing a collection of materials together and placing the resultant composition in a form. When properly designed, consolidated, and cured, PCC (with adequate reinforcing steel) will provide excellent structural properties in the field. Improper design, preparation or placement 1

12 can yield an inferior quality PCC of low strength and high porosity. Poor quality control can produce various forms of deterioration. The main causes of deterioration in PCC are chloride-induced corrosion of the reinforcing steel, freeze-thaw damage, and alkali-silica reaction. Chlorides found in PCC structures, often come from the use of salts as a deicing agent on roads, their presence in spray areas near salt-water bodies, and their inclusion during mixing. Freeze-thaw damage in PCC results from the expansion of moisture caused by freezing temperatures. This cyclic process exerts destructive tensile pressures within the PCC. Alkali-silica reaction occurs when an aggressive reaction takes place between alkalis in the PCC pore water and silica ions (from amorphous, high-silica content aggregate). This also causes tensile pressures to build within the PCC. Generally, the aforementioned deleterious factors are active beneath the PCC s surface and cannot be accurately assessed by visual observation. A majority of repair and rehabilitation funds are therefore used to fix conditions unseen until the work is contracted and repair work begins or until the deterioration is visible, because it has reached such an advanced stage that it may hinder the use and function of the structure. Recognizing the potential for damage before it occurs will help to preserve the facility s structural integrity, reduce life-cycle costs, and minimize the disturbance to facility users. To asses the physical condition of large PCC structures without causing further damage, nondestructive evaluation (NDE) methods have been developed. Their importance results from the noninvasive nature of the techniques used and the anticipated rapidity of the measurements. However, not all NDE methods have been welcomed (or understood) by many practicing civil engineers, who do not have the interdisciplinary background or the inclination to learn how to properly transform noninvasive methods into an operational tool. Consequently, although the concept of a noninvasive measurement technique is attractive these engineers, there is a gap between laboratory concept and field application. Portland cement concrete is a composite containing a variety of materials each with different electrical properties. However, electromagnetic (EM) characterization of 2

13 PCC using NDE methods can be used to impart information about its constituent materials, thereby revealing information about the properties of the composite itself. The EM properties of interest in this regard are conductivity and relative permittivity (dielectric constant). These electrical properties are related to the composite properties of the aggregate, aggregate size, cast water to cement (w/c) ratio, chloride content, and moisture content. As chemical and physical changes occur within PCC due to deterioration, the local and bulk EM material properties and the propagation of EM waves in the material also change. The dielectric properties of PCC have been investigated in a laboratory setting at Virginia Tech using a parallel-plate capacitor operating in the frequency range of MHz (Al-Qadi and Riad, 1996). This capacitor setup consists of two horizontalparallel plates with an adjustable separation for insertion of a dielectric specimen (e.g., PCC). The parallel-plate capacitor has shown that detection of different types of deterioration in PCC is feasible, but it is frequency dependent. Chloride contamination, for example, can be easily detected at low radio frequencies ( MHz), but is difficult to detect at low microwave frequencies (1-10 GHz). The parallel-plate results were correlated to the chloride content and different prediction models were established (Al-Qadi et al., 1997). 1.2 Problem Statement Civil engineering structures constructed with PCC deteriorate with time and need to be maintained or replaced. Preventative maintenance procedures can often reduce the life-cycle cost of such structures. Techniques that would allow civil engineers to detect subsurface deterioration during routine inspections would aid in their efforts to prevent major defects from occurring. However, by the time such deterioration becomes evident on the surface, it is often too late to apply low-cost maintenance procedures. Early detection and evaluation of deterioration in PCC (e.g., alkali-silica reaction, freeze-thaw damage or chloride presence) would allow engineers to optimize the life-cycle cost of a constructed facility and minimize disturbance to its users. 3

14 While useful in research, the parallel-plate capacitor is not practical for field implementation. Therefore, a new instrument, based on the parallel-plate capacitor, needs to be developed for civil engineers to detect internal deterioration and to provide field measurements of the dielectric properties of PCC. 1.3 Objectives of Research The objective of this research is to develop a capacitor probe that yields measurement results comparable to those of the parallel-plate capacitor. Since the parallel-plate capacitor has proved to give accurate and reliable measurements of the dielectric properties of materials, it has been chosen as the standard measurement device to which the capacitor probe will be compared. Since it is also desirable to produce a capacitor probe that is reusable for multiple measurement events, it is to be an in-situ probe that is lightweight, flexible, durable, and inexpensive. 1.4 Scope of Research To achieve the objectives of this study, a capacitor probe was developed to measure the effect of different parameters on the complex dielectric constant of PCC over low radio frequencies (2-20 MHz). This was accomplished by preparing PCC mixes with different PCC parameters such as air content. Control mixes were prepared using a water to cement (w/c) ratio of 0.45 and air contents of 2 and 6%. Deterioration was induced in some specimens prepared at a 0.45 w/c ratio and 6% air content by inserting a Styrofoam layer during the casting process. The effect of PCC maturity was studied by evaluating the dielectric properties at different curing times. Chapter 2 sets out the physical and chemical properties of PCC, mechanisms of common forms of deterioration, and basic dielectric theory. Chapter 3 describes the parallel-plate capacitor measurement system and the theoretical equations governing its operation. Chapter 4 presents the newly developed capacitor probe measurement system and the theoretical equations regarding its usage. Chapter 5 yields the 4

15 experimental program involved in this research experiment. Chapter 6 presents and discusses the results of the experimentation. Chapter 7 offers the summary, findings, and conclusions, and Chapter 8 makes recommendations for further studies. 5

16 CHAPTER 2 BACKGROUND Current NDE methods typically use EM waves as a vehicle to gather information about the material under examination. One of the earliest adaptations of EM technology to civil engineering involves EM pulse radar. Although originally used for geological exploration (Lundien, 1971; Hipp, 1971; Ellerbruch, 1974; Feng and Delaney, 1974; Moffat and Puskar, 1976; Lord et al., 1979; McNeill, 1980; Shih and Doolittle, 1994; Feng and Sen, 1985; and Shih and Myhre, 1994), this technique has been used for highway and bridge applications (Steinway et al., 1981; Clemena, 1983; Carter et al., 1986; Clemena et al., 1986; Chung and Carter, 1989; Eckrose, 1989; Bungey and Millard, 1993; and Maser, 1996). Pavement condition studies have been conducted in which subsurface voids were detected (Steinway et al., 1981; Clemena et al., 1987) and layer thickness and subsurface moisture measurements were performed (Bell et al., 1963; Maser et al., 1989; Al-Qadi et al., 1989). Additionally, dielectric properties have been measured to determine the moisture content of soils (Topp et al., 1984; Dobson and Ulaby, 1986; Jackson, 1990; Campbell, 1990; Scott and Smith, 1992; Brisco et al., 1992; and Straub, 1994). Electromagnetic waves have been used for agricultural applications (Nelson, 1985) and for measuring the dielectric properties of food items (Bodakian and Hart, 1994). Liu et al. (1994) and Steeman et al. (1994) measured dielectric properties of various materials for characterization in the electronics industry. Before PCC can be characterized successfully, the electromagnetic properties of the constituents of this composite material must be understood. In this regard, McCarter and Curran (1984) have demonstrated that characteristics of the electrical response of cement paste could be used as an effective means for studying the progress of hydration and structural changes occurring within cement paste. Taylor and Arulanandan (1974) have also investigated the relationship of mechanical and electrical properties (conductivity and capacitance) of cement pastes measured at early ages over a frequency range of 1 to 100 MHz. Whittington and Wilson (1986) have 6

17 researched the effect of curing time on the conductivity of PCC and its relationship with compressive strength. A detailed discussion of the hydration of Portland and non- Portland cements within the first 24 hours with respect to conductivity has been presented by Tamas et al. (1987) and Perez-Pena et al. (1989). They discussed the influence on conductivity due to accelerators and retarders (from 2 Hz to 2 MHz) and the effect of inorganic admixtures (chlorides and hydroxides at 1, 10, 100, and 1000 khz). Wilson and Whittington (1990) have presented the relationships between conductivity of PCC and frequency during early stages of curing. The electrical resistivity of PCC has also been investigated. Whittington, McCarter, and Forde (1981) have compared the measured resistivity values for PCC of varying composition with a theoretical model. A new technique for observing the time dependent resistivity measurements of PCC with varying compositions has been developed by Hansson and Hansson (1983) and compared with a theoretical model. Wilson and Whittington (1990) have discussed the validity of a developed theoretical model which describes the frequency based (1 to 100 MHz) dependence of the resistivity of PCC. Similarly, De Loor (1962), Wittmann and Schlude (1975), Perez-Pena et al. (1989), and Moukwa et al. (1991) have studied the dielectric properties of PCP over the RF and microwave frequencies. Whittington et al. (1981), McCarter and Whittington (1981), Hansson and Hansson (1983), McCarter and Curran (1984), McCarter et al. (1985), and Wilson and Whittington (1990) have performed measurements of dielectric properties of PCC over RF. Hasted and Shah (1964) and Shah et al. (1965) have measured the dielectric properties of bricks, Portland cement, and PCC at different w/c ratios. Results are compared to theoretically obtained values. Dielectric properties of PCC have been measured in an effort to gain a better understanding of its mechanical properties. However, the basic relationships between electromagnetic and mechanical properties of PCC structures are not always well understood. These properties have been measured using three different systems over a wideband frequency: a parallel-plate capacitor, a coaxial transmission line, and TEM Horn antenna. Al-Qadi et al. (1994b, 1995, and 1997), Al-Qadi and Riad (1996), and Haddad (1996) have presented the development and use of a parallel-plate capacitor 7

18 operating in the low radio frequency range (0.1 to 40.1 MHz). Al-Qadi et al. (1994a and 1995) and Al-Qadi and Riad (1996) have developed a coaxial transmission line fixture that operates over a frequency range of 0.1 to 1 GHz. Ghodgaonkar et al. (1989) have developed a microwave measurement fixture employing an antenna to determine the dielectric constant at a frequency range of 14.5 to 17.5 GHz. While Al-Qadi et al. (1991) have implemented a new setup to measure the dielectric constant of hot-mix asphalt at GHz. Al-Qadi et al. (1996) and Al-Qadi and Riad (1996) describe another antenna fixture developed to measure the influences of induced deterioration on the dielectric properties of PCC slabs from 1 to 10 GHz. Tewary et al. (1991) presents theory regarding a non-contact system to measure the capacitance of materials. This capacitance probe was modeled to operate in a manner similar to a parallel-plate capacitor, except that the plates lie within the same horizontal plane. A similar but expanded method was developed (Diefenderfer et al., 1997; Yoho, 1998) creating a surface probe to measure in-situ dielectric properties of PCC. An overview of PCC and dielectric measurements follows. 2.1 Portland Cement Concrete Portland cement concrete is a composite material consisting of cement, water, and coarse and fine aggregate. While anhydrous Portland cement does not posses any bonding properties, its union with water allows it to act as an adhesive to unite these materials into a cementitious composite. Upon inception of this hydration reaction, PCC begins to harden and hydration products are formed. Unlike most other construction materials, PCC is a dynamic system. Some components of PCC continue to gain strength with time; in fact, the word concrete is derived from the Latin term concresure meaning to grow together (Lewis and Short, 1907). The hydration process of cement determines the internal structure of PCC. The type of cement, stage of hydration, curing and temperature conditions, and the proportions of the mixture ingredients define PCC's final internal structure. Although the aggregate is often considered filler in ordinary strength PCC, it plays an important 8

19 role in determining the mixture s durability. For the purpose of examination, PCC can be broken down into three distinctly different parts: hydrated cement paste, aggregate, and transition zone which is located between the cement paste and the aggregate Hydration of Portland Cement The main ingredients used to produce Portland cement are lime, silica, alumina, and iron oxide. These materials react in a kiln during the production of Portland cement to form more complex compounds, the main components are abbreviated by civil engineers as C 3 S (Tricalcium silicate), C 2 S (Dicalcium silicate), C 3 A (Tricalcium aluminate), and C 4 AF (Tetracalcium aluminoferrite) where C = CaO, S = SiO 2, A = Al 2 O 3, and F = Fe 2 O 3. Minor constituents include MgO, TiO 2, Mn 2 O 3, K 2 O, and Na 2 O. Their listing as minor describes only their relative quantity (a small percent of the weight of the Portland cement) and not their importance in the PCC mixture. Of particular interest to civil engineers are sodium and potassium oxides, which have been found to react with certain types of aggregates. The products of this reaction have been shown to cause disintegration in PCC (Neville, 1981). The processes by which Portland cement and water form a bonding substance take place in a water-cement paste. That is, in the presence of water, the silicates and aluminates listed above react to form products of hydration. The principal cementing agent (comprising approximately 50-60% of the total solid volume of hydration products) is calcium silicate hydrate (C-S-H). A material of poor crystalline structure, C- S-H is made up of an extremely fine (less than 1 µm) conglomeration of calcium silicate hydrate and other crystallites formed as a result of the hydration of Portland cement. The other main hydration product (comprising approximately 20-25% of the total solid volume of hydration products) is calcium hydroxide. Calcium hydroxide, thought to be much less cementitious than C-S-H, probably adds little to the cementitious properties of the final mixture. With time, these products of hydration become the hardened cement paste. This hydration process begins at a rapid rate that then decreases with time. If maintained at 100% relative humidity, approximately 75% of the cement hydrates within the first 28 days. However, the process has been noted to continue for 9

20 up to 50 years and is believed to never completely cease should water be present (Taylor and Arulanandan, 1974). The equations denoting the chemical reactions involved in the hydration process were developed assuming that each reaction is independent of the others. While this is not entirely true, it does provide an accurate assessment of the reaction process. The following equations give a simplified view of the reactions involved in the hydration process (Mindess and Young, 1981): 2C 3 S + 6H C 3 S 2 H 3 + 3CH (2.1) 2C 2 S + 4H C 3 S 2 H 3 + CH (2.2) C 3 A + 3CSH H C 6 AS 3 H 32 2C 3 A + C 6 AS 3 H H 3C 4 ASH 12 C 4 AF + 3CSH H C 6 (A,F)S 3 H 32 + (A, F)H 3 C 4 AF + C 6 (A,F)S 3 H H 3C 4 (A,F)SH 12 + (A,F)H 3 (2.3a) (2.3b) (2.4a) (2.4b) where H = Water (H 2 O); C 3 S 2 H 3 = Calcium hydrate silicate (C-S-H); CH = Calcium hydroxide; CSH 2 = Gypsum; C 6 AS 3 H 32 = 6-calcium aluminate trisulfate-32-hydrate (Ettringite); 3C 4 ASH 12 = Tetracalcium aluminate monosulfate-12-hydrate (monosulfoaluinate); and 3C 4 ASH 12 = tetracalcium aluminate monosulfate-12-hydrate. 10

21 Calcium SIlicates The first of five reaction stages describing the hydration of calcium silicates in Portland cement is defined by a rapid evolution of heat; it lasts for only a few minutes after mixing water with Portland cement. The hydrolysis of C 3 S (Equation 1) begins quickly and releases both calcium and hydroxide ions into solution. The ph of the mixture rises above 12, indicating high alkalinity. The first product formed in the hydration reaction, calcium hydrate silicate (C-S-H) gel, has a CaO:SiO 2 ratio of nearly 3. This is identical to the molar ratio in the anhydrous C 3 S compound. Dicalcium silicate (C 2 S) will hydrate in a simillar manner to tricalcium silicate (C 3 S); however, C 2 S is much less reactive and, therefore, less heat is evolved during this process (Equation 2). Stage 2 begins when calcium silicate hydrate begins to coat the remaining C 3 S and retards further hydration. This action also marks a dormant period of little hydration activity, which temporarily keeps PCC in a plastic state. This temporary halt to hydration is needed to achieve a certain concentration of ions in solution before the next hydration products can form from crystal nuclei. The initial products of hydration, however, are unstable and begin to crystallize from solution when calcium and hydroxide concentrations reach a critical value. This results in an accelerated reaction involving C 3 S, marking the onset of stage three. During the third stage of the hydration reaction, another hydration product is formed: C-S-H (I) gel with a CaO:SiO 2 ratio of 1.5 or less. This is followed immediately by C-S-H (II) gel with a CaO:SiO 2 ratio of 1.5 to 2.0. During stage three, the rate of heat evolution increases to a peak at approximately 6 to 11 hrs after the onset of hydration. Calcium silicate hydrate continues to coat the C 3 S grains in an everthickening barrier. Therefore, water is only able to penetrate the anhydrous C 3 S grains through diffusion. Stage four is marked by both chemical- and diffusion-controlled rates of hydration involving the coated C 3 S. Eventually, this process is totally controlled by the rate of diffusion, thus marking the onset of the diffusion-controlled stage five. Hydration in this manner is quite slow and approaches 100% hydration asymptotically. 11

22 Tricalcium Aluminate The hydration of C 3 A involves reactions with sulfate ions supplied by gypsum (CSH 2 ). Gypsum is added to the vitrified cement (also called clinker) during production in the kiln to prevent flash setting, which is an immediate stiffening of cement paste due to the reaction of C 3 A and water. The hydration of C 3 A, described in Equation 2.3a, produces a material known as ettringite. The formation of ettringite slows the hydration of C 3 A by forming a diffusion barrier around the unhydrated grains, much as C-S-H slows the hydration of calcium silicates. Ettringite is a stable product so long as sufficient sulfate is available (supplied in the form of gypsum). Once all available sulfate is consumed, ettringite transforms into monosulfoaluminate (Equation 2.3b), another calcium sulfoaluminte hydrate which contains less sulfate. This transformation again allows C 3 A to rapidly react with water and typically occurs within 12 to 36 hrs after all gypsum has been used to form ettringite. The initial heat release observed within approximately five minutes of adding water to cement occurs because the hydration retardation properties of gypsum have yet to begin. Tricalcium aluminate is undesirable in PCC, because when hardened cement paste is attacked by sulphates, formation of calcium sulphoaluminate from C 3 A causes expansion in the hardened paste. Adding little or nothing to the strength of cement except at early ages, C 3 A acts as a flux to reduce the temperature required to burn the clinker during the manufacture of Portland cement and aids in the combination of lime and silica. Ferrite Phase Tetracalcium alumino-ferrite (C 4 AF) and tricalcium aluminate (C 3 A) form similar sequences of hydration products. However, C 4 AF can form these with or without the presence of gypsum. Additionally, the reactions involving C 4 AF (Equations 2.4a and 2.4b) are much slower and produces less heat than the hydration reactions of C 3 A. Iron oxide which modifies the rate of hydration, can be substituted for alumina (as seen from the compound C 6 (A,F)S 3 H 32 ) with little change in the hydration reaction. 12

23 2.1.2 Hydrated Cement Paste In addition to the solid phase described above (including unhydrated clinker), hydrated cement paste (HCP) is also comprised of void spaces that may contain water in various forms. These voids are neither uniform in size nor uniformly distributed throughout the paste. Voids in Hydrated Cement Paste On average, the bulk density of the products of hydration is less than the bulk density of the anhydrous Portland cement. It has been estimated that one cm 3 of Portland cement occupies approximately two cm 3 of space after complete hydration (Mehta and Monteiro, 1993). Various types of voids, which account for this differing bulk density, are present in hydrated cement paste (HCP), interlayer space in C-S-H, capillary voids, and air voids. The interlayer space in C-S-H is considered to be approximately 5-25 Å in size. While this void space is too small to have any adverse effect on strength, it may contain enough water by hydrogen bonding that its release (by breaking the hydrogen bond) into the capillary void structure may contribute to drying shrinkage and creep. The volume and size of the capillary voids is determined by the cast w/c ratio and the degree of hydration. The total volume of the capillary voids is calculated by determining the porosity of the mixture. Capillary voids may vary in size from nm for well-hydrated pastes with a low w/c ratio to 3-5 µm for high w/c ratio pastes at early ages. Macropores, capillary voids larger than 50 nm, are thought to play a role in determining strength and permeability. Micropores, capillary voids smaller than 50 nm, are thought to affect drying shrinkage and creep. It is assumed that the capillary pores form an interconnected network that can be assessed by fluids and gases that penetrate and permeate the concrete. The main mechanism of fluid transport in PCC is due to capillary forces and hydrostatic pressure. Gaseous flow is attributed to partial pressure gradients and external pressures. The ease with which fluids and gases can pass through PCC increases with increasing porosity of the PCC. The w/c ratio is an important factor in determining the porosity of 13

24 PCC. At the same degree of hydration, a low w/c ratio mixture produces fewer pores of smaller size than a high w/c ratio mixture. Typically spherical in shape, air voids are the largest voids in the HCP. Air voids can result from air entrainment or air which is entrapped during the mixing process. Entrapped air voids are approximately up to 3 mm in diameter, while entrained air voids range from 50 to 200 µm in diameter. Entrapped and entrained air, being larger than capillary voids, can affect the strength and permeability of the concrete mixture Aggregate Phase Although usually considered an inert filler, aggregate (including both fine and coarse) makes up nearly 60 to 80%, by volume, of PCC. The aggregate is predominately responsible for the unit weight, elastic modulus, and dimensional stability of the concrete mixture (Mehta and Monteiro, 1993). These properties are determined not by their chemical composition but by their physical attributes, such as volume, size, pore distribution within the aggregate, shape, and texture. These attributes are derived from the parent rock, exposure conditions, and processes used to manufacture the aggregate. Aggregate can be divided into two classes by size and two classes by weight. Coarse aggregate is typically larger than 4.75 mm, while fine aggregate is generally between 4.75 mm and 75 µm in size. Normal weight concrete (approximately 2400 kg/m 3 ) can be made using aggregate with a bulk density of kg/m 3. Lightweight and heavyweight concrete can be made using aggregate with a bulk density less than 1120 kg/m 3, and from aggregate with a bulk density greater than 2080 kg/m 3, respectively. Many factors can influence the effect of aggregate on PCC, including the maximum size, coarse/fine aggregate ratio, shape, texture, and material composition. Concrete with a larger maximum aggregate size requires less mixing water than concrete with a smaller maximum aggregate size. The former generally leads to stronger concrete; however, larger aggregate tends to have weaker transition zones. 14

25 The net effect of these two tendancies is a function of the w/c ratio of the PCC and the applied stress. If the maximum aggregate size and the w/c ratio are kept constant and the coarse/fine aggregate ratio is increased, the strength usually decreases. Crushed aggregate is usually stronger in tension than naturally weathered gravel of the same mineralology. Also, it is assumed that a stronger mechanical bond between the aggregate and the cement paste exists. This bond preference is more pronounced at early ages. Although crushed aggregate may be stronger than smoother gravel, more mixing water is required to achieve the same workability when using more roughly textured aggregate. This may offset any advantages gained by aggregate texture Transition Zone As the weakest link of the chain, the transition zone is considered the strengthlimiting phase of PCC. The transition zone exists between large particles of aggregate and the HCP. Even though it is composed of the same components that exist in the HCP, the properties of the transition zone differ from the HCP. This difference is seen as the transition zone fails at a much lower stress level than either of the two main components of PCC. In fact, 40-70% of the ultimate strength is a large enough quantity to extend cracks already present in the transition zone. At 70% of the ultimate strength, stress levels are sufficiently high to initiate cracking in large voids in the HCP. As stresses increase beyond this level, the cracks will begin to extend from the HCP to the transition zone. This makes the crack continuous and, thus, ruptures the material. While it is difficult to extend cracks in PCC under compressive loading, it is relatively easy to extend cracks under tensile loading. This, in part, explains why PCC is much weaker in tension than in compression. Adhesion between the hydration products and aggregate particles is due to Van der Waals forces of attraction. Therefore, the strength of the transition zone is dependent upon the size and volume of voids present. At early stages, the size and volume of voids in the transition zone are larger than in the bulk HCP. Consequently, the strength of the transition zone is lower than the bulk HCP. However, with time, the 15

26 transition zone becomes nearly as strong as the bulk HCP. It is assumed that this occurs due to the formation of new products in the void spaces by slow reactions between the constituents of the cement paste and the aggregate. These reactions also reduce the amount of the less-adhesive calcium hydroxide. Additionally, microcracks help to weaken the transition zone. The number of microcracks is a function of the aggregate size and grading, cement content, w/c ratio, degree of consolidation, curing conditions, environmental humidity, thermal history of the concrete mixture, impact loads, drying shrinkage, and sustained loads at high stress levels (Mehta and Monteiro, 1993). However, some microcracks are present even before the finished structure is loaded. 2.2 Deterioration of PCC Generally, deterioration of PCC takes place involving one or more of the constituents of PCC and aggressive reactants from the external environment. Among these forms of deterioration are the electrochemical corrosion of embedded steel due to chloride intrusion into PCC, carbonation, alkali-silica reaction (ASR), and freeze-thaw damage. Deterioration often begins as a chemical reaction but results in physical defects such as increased porosity and permeability, decreased strength, and/or cracking and spalling. In-situ PCC structures may experience several chemical and physical deterioration processes simultaneously; in fact, some may accelerate the effects of others. A brief description of three forms of deterioration follows Corrosion of Reinforcing Steel in PCC Bradford (1992) states that the direct cost of structural deterioration due to corrosion of reinforcing steel in an industrialized nation consumes approximately 4.9% of the gross national product of that nation. Metallic reinforcing is used in PCC structures for several reasons, one of which is because the failure of metallic reinforced PCC is less brittle than the failure of unreinforced PCC. Metallic reinforcement placed in PCC structures is usually concentrated in areas of greatest tensile forces. If the 16

27 structure were to fail due to these tensile forces, the metallic reinforcement would stretch to a certain degree due to its high ductility. However, an unreinforced PCC structure would undergo catastrophic brittle failure. Steel is most often used as metallic reinforcement in PCC since the coefficients of thermal expansion for steel and PCC are similar. Properly mixed and placed PCC (having a sufficient cover depth) usually provides adequate protection for internal reinforcing. In addition, the high ph of PCC (typically ) provides an environment in which the oxides of iron are thermodynamically stable. A passive protective film of iron oxide is created around the steel reinforcement in the presence of water, oxygen, and water-soluble alkaline products (predominately calcium hydroxide) from the hydration of cement. This film of corrosion products slows the rate at which further corrosion can occur and protects the remaining metal from further corrosion. In this passive state, steel corrodes at a rate approximately equal to 10 x 10-6 cm/yr (Hansson and Sorensen, 1990). However, if this film becomes soluble, the passivity of the steel is eliminated and corrosion continues at an increased rate. Two ways, in which the passive film layer is destroyed, are a reduction in the alkalinity of the concrete and an electrochemical reaction involving chloride ions in the presence of oxygen. The alkalinity can be reduced by the leaching of alkaline substances with water or by a neutralization effect involving carbon dioxide. Chlorides are often present due to use of salts as a deicer on roads, spray from seawater, and inclusion during mixing. However, reinforcing steel is usually covered by approximately 25 mm of PCC (at least when properly constructed). Therefore, these deleterious results only occur when the corrosion-causing agents reach the steel after penetrating the PCC. The electrochemical process describing the corrosion of reinforcing steel in PCC involves an anode (site of electrochemical reduction), a cathode (site of oxidation), an electrolyte (in PCC, the paste-pore solution), and an electrically continuous connection. The removal of any one of these four components will halt the corrosion reaction. The anode and cathode sites develop as corrosion cells of differing electrochemical potential, where the reaction occurring at the cathode consumes electrons produced by 17

28 a reaction at the anode. Electrochemical potentials can form from the presence of dissimilar metals (e.g., steel reinforcing and aluminum conduit) and differences in concentrations of dissolved ions (alkalis, chlorides, and oxygen) near the reinforcing steel. One of the most common causes of corrosion of reinforcing steel in PCC is the presence of chlorides (Rosenberg et al., 1989). The proper amounts of oxygen and moisture in close proximity to the internal reinforcing steel, combined with chlorides, can lead to deterioration (ultimately by delamination and spalling) of the PCC structure. According to Al-Qadi et al. (1993) reinforcing steel will begin to corrode when the concentration of chloride ions in the pore solution reaches a threshold level of 0.6 kg/m 3 of PCC. Chlorides, in the form of salt-contaminated aggregate, deicing salts, or seawater can penetrate PCC structures through cracks or diffusion through the PCC s pore water. Free chloride ions (Cl - ) moving through the PCC pore system react with Fe 2+ in areas where the passive coating (γ-fe 2 O 3 ) surrounding the embedded steel has been destroyed. This passive layer is reported to be stable when the ph of the pore solution remains above 11.5 (Mehta and Monteiro, 1993). Additional Cl - and H + ions are released and iron hydroxide is formed when FeCl 2 undergoes further reactions in the presence of moisture. The newly-formed iron hydroxide reacts with oxygen to form Fe 2 O 3 (rust). This transformation of metallic iron to rust is also accompanied by an increase in volume, potentially occupying up to six times the original metallic volume. This process results in a reduction of the effective area of the reinforcing steel and creation of tensile forces in the PCC structure; ultimately, this processs can lead to cracking, delamination (between the reinforcing steel and the surrounding PCC), and spalling. Cracks that extend to the surface of the PCC allow more chlorides to intrude into the PCC and thus perpetuate the reaction. There are many factors that will influence and control the rate at which the corrosion of reinforcing bars occurs. Higher temperatures will increase the rate of corrosion by up to two times for every 10 C increase in temperature. A higher moisture content in the concrete will also increase the rate of corrosion by providing an electrolyte for the transfer of electrons from the anode to the cathode. However, 18

29 oxygen can diffuse more easily into dry concrete than wet concrete, because diffusion through water is much slower than diffusion through air Alkali-Silica Reaction Alkali-silica reaction (ASR) is an expansive chemical reaction involving alkali ions present in the PCC paste and certain siliceous materials present in the aggregate. The high ph level present in the PCC pore solution results in a highly alkaline solution in which aggregates formed from silica and siliceous materials do not remain stable after long periods of exposure. Ultimately, this expansive reaction can lead to pop-outs and exudation of an alkali-silicate fluid. Alkali-silica reaction was first recognized as a problem in the United States in the New River Valley area of Virginia in the 1930s (Hobbs, 1988). In order for this expansive reaction to occur, both hydroxyl ions and alkali-metal ions are necessary. Hydroxyl ions are present in hydrated Portland cement due to the existance of calcium hydroxide; therefore, the amount of alkali-metal ions will control the degree of ASR reaction. These alkali-metal ions are introduced into PCC through alkali-containing admixtures, salt-contaminated aggregates or penetration of seawater or deicing solutions containing sodium chloride. Alkaline hydroxides, derived from alkalis (Na 2 O and K 2 O), attack siliceous material in the aggregate. This process results in the formation of an alkali-silicate gel that swells in size due to the osmosis of water. Since this gel is confined within the surrounding cement paste, destructive tensile forces develop within the PCC. Two theories exist that may explain the mechanism of expansion caused by ASR (Hobbs, 1988). One theory attributes the stresses generated within the PCC to expansion of the ASR gel by absorption of the pore fluid. The other theory attributes the induced stresses to an osmotic pressure generated across impermeable memebranes. According to the absorption theory, expansion depends on the volume concentration, growth rate, and physical properties of the alkali-cement gel. Additionally, the amount of damage is proportional to the rate of gel growth. At a high 19