26 CHAPTER 3: ELECTRICAL MEASUREMENTS IN CONCRETE MATERIALS This chapter provides a review of the history and application of electrical measurements in concrete. The definition of electrical conductivity and the principles of direct (DC) and alternating (AC) conductivity measurements are presented. The theoretical principles relating the conductivity of a porous media to its microstructural properties and humidity condition is discussed. 3.1. A Brief Introduction to Electrical Measurements in Concrete In concrete materials research, electrical measurements (EM) measure the opposition of the material to flow of electricity. For example, one can apply a direct (DC) voltage to the ends of a concrete sample, and measure the sample’s electrical resistance. Most early researchers, however, recognized that DC measurements may not be accurate due to polarization effects. For example, Hammond and Robson (1955) observed a change in the measured resistance when they reversed the direction of electrical current. For this reason, EM methods which use alternating current (AC) have become more applicable. Hundreds of research studies have used electrical measurements to investigate different phenomena in concrete materials. Christensen and coworkers (1994) used EM to perform an in depth investigation on microstructural changes in hydrating cement concrete. They observed an increase in the measured resistance with age of the sample and concluded that this increase is mainly attributed to the consumption of the liquid phase of the matrix and production of large volumes of solid products during hydration process. They showed that EM can provide valuable information about concrete’s porous 27 microstructure. Also measurements can be related to factors such as water permeability and ionic diffusivity of the matrix. Electrical measurements have been used to monitor moisture transport in concrete. McCarter and coworkers observed a change in the resistance value when a wet sample is oven-dried (McCarter and Garvin 1989). Several researchers have studied water and ionic penetration in concrete’s cover zone and used EM to develop and assess moisture profiles in drying concrete (McCarter et al. 1996, McCarter 1996, McCarter and Watson 1997, Weiss et al. 1999, and Schieβl et al. 2000). EM has also been widely used to measure corrosion potentials in reinforced concrete (examples include ASTM C876 and the study by Schieβl and Breit 1995). Electrical methods provide several advantages for accurate in-situ testing of materials. These methods are non-invasive and non-destructive and do not necessitate water removal from samples prior to testing. Yet, there are also some disadvantages involved. It is still unclear how to differentiate between separate underlying factors responsible for electrical conduction in concrete: Factors such as different ions dissolved in the pore solution, their concentrations, water content, and geometry of the microstructure. Also a widely accepted standard test procedure does not exist. Each person tends to develop his/her own testing protocols, and as a consequence, it is difficult to compare results or replicate tests from different laboratories (McCarter, 2002). 3.2. Resistivity and Conductivity Ohm’s law in its simplest form relates a DC electrical current (I), passing through a conductive sample, to the voltage applied to the sample’s ends; V = RI . In this equation R is electrical resistance which is a function of a material property (so called resistivity), geometry, and dimensions of the sample. Figure 3.1 shows a cylindrical specimen under constant DC voltage. Measuring resistance of the samples with different dimensions shows that electrical resistance is proportional to the sample’s length (L) and inverse to the surface area (Equation 3.1): 28 R = mρ L A (3.1) when ρ is the material’s resistivity and m is a geometric factor which equals to “one” (m=1) for a cylindrical sample with disc shape electrodes at the ends (Figure 3.1). As a result, resistivity of the material can be determined by measuring the sample’s electrical resistance and multiplying the measured value by a factor representing the dimensions and geometry of the test setup ( ρ= A R mL or ρ= 1 in Equation 3.2): k 1 R k (3.2) geometry factor Conductivity (σ) is defined as the inverse of resistivity (Equation 3.3): σ= 1 ρ = k R (3.3) Conductivity (or resistivity) is a geometry free parameter that represents the opposition of the material to flow of electricity. Conductivity of concrete, as will be discussed later in this chapter, is a function of the microstructure’s geometry, moisture content of the sample, and conductivity of the pore solution. A Fig. 3.1. A cylindrical sample under constant DC voltage can be used to see how dimension and geometry influence the measured electrical resistance 29 The procedure of measuring the material’s conductivity is slightly different when alternating (AC) currents are used. The equivalent form of Ohm’s law for AC currents is V = ZI , in which Z is electrical impedance and is again a form of opposition to flow of electricity. Electrical impedance is a complex number composed of real and imaginary components; Z = Z ′ + jZ ′′. The real part of impedance ( Z ′ ) is called resistance, while the imaginary part ( Z ′′ ) is called reactance. In AC measurements, the sign and magnitude voltage is not constant. For example the voltage can oscillate between positive and negative peaks in a sinusoidal manner with constant frequency. The real and imaginary components of electrical impedance are both functions of this frequency. In fact, resistance and reactance of a sample can be measured at different frequencies and plotted against each other. Such a graph is called Nyquist plot (Figure 3.2a). Each point in this graph corresponds to a distinct frequency. The impedance components ( Z ′ and Z ′′ ) can be combined to represent the total (or absolute) impedance, Z , and the impedance argument (or phase angle), θ : Z = ( Z ′) 2 + ( Z ′′) 2 (3.4) ⎛ Z ′′ 180 o ⎞ ⎟⎟ φ = tan ⎜⎜ − ⋅ ⎝ Z′ π ⎠ (3.5) −1 The values of Z and θ for each point can be plotted against the voltage’s frequency to get Bode plots (Figure 3.2b). A typical Nyquist plot for concrete is shown in Figure 3.3. This graph is composed of two half arcs, a larger arc, which is typically in the millihertz to kilohertz range and is attributed to the passive oxide film on the surface of steel electrodes, and a smaller arc, typically measured in the kilohertz to megahertz range and is attributed to the sample’s bulk properties. 30 -100000 Z'' increase in frequency -50000 |Z| θ 0 -50000 0 50000 100000 150000 Z' (a) 106 |Z| 105 104 103 102 103 104 105 106 107 106 107 Frequency (Hz) -100 theta -75 -50 -25 0 102 103 104 105 Frequency (Hz) (b) Fig. 3.2. (a) Nyquist plot, and (b) Bode plots of electrical impedance 31 Fig. 3.3. Typical Nyquist plot for hardened concrete (Christensen et al. 1994) One of the most important parameters that can be obtained from the Nyquist plot is the “bulk resistance” ( Rb ) which is the resistance ( Z ′ ) at the intersection of the two arcs. At this point the reactance ( Z ′′ ) of the system is theoretically zero. The frequency corresponds to this point is called the “cut-off” frequency. The measured bulk resistance is converted to concrete’s conductivity using Equation 3.3. In the past, single frequency AC measurements were made at a frequency in the order of kilohertz to measure the bulk resistance. Christensen and coworkers (1994) showed that that the cut-off frequency can vary over two orders of magnitude as a function of hydration time, cement type, and w/c ratio. Weiss et al. (1999) showed that this frequency also varies as a function of the moisture content of concrete. As a result, in order to accurately measure the bulk resistance, it is necessary to measure impedance over a wide range of frequencies. Such a process is called impedance spectroscopy (IS). Impedance spectroscopy can be performed by employing a gain-phase analyzer and a personal computer for data acquisition. 32 3.3. Conductivity of Cementitious Matrices According to the Powers-Brownyard model (Taylor 1990), cement paste can be envisioned as parallel layers of different components (i.e., capillary porosity, hydration products, unhydrated cement, etc.) that extend from one side of a sample to the other (Figure 3.4). The conductivity of a set of n parallel layers can be calculated using the followings approach. The complete electrical resistance of the set of n layers (Rt) is related to the resistance of each ith layer (as represented by Ri): 1 1 1 = +"+ Rt R1 Rn (3.6) Rewriting this expression in terms of conductivity (si and Ai are respectively conductivity and cross sectional area of the ith layer, and m is the geometry factor that is the same for all layers) results in: Ri = σ i Ai (3.7) mL σ t At = σ 1 A1 + " + σ n An Direction of current (3.8) 1 Surface area= A1 2 I . . . I n Length= L Fig. 3.4. Powers-Brownyard model of envisioning cement paste as parallel layers 33 Multiplying both sides by the length (L) gives: σ tVt = σ 1V1 + " + σ nVn (3.9) when Vi is the volume of layer i. When the volume fraction of layer i is defined as follows: φi = Vi Vt (3.10) Equation 3.9 can be rewritten in terms of volume fraction of layers: σ t = σ 1φ1 + " + σ nφ n (3.11) If the layers were in series (Figure 3.5), Equation 3.11 would get the following form: σt = σ σ1 +" + n φ1 φn (3.12) Since in an actual concrete sample, the constituting components form a complex system of serial and parallel layers, none of the Equations 3.11 and 3.12 can be exclusively used. To consider this fact, Christensen (1993) used a modified equation in the following form: σ t = σ 1φ1 β 1 + " + σ nφ n β n (3.13) Length= L1 I 1 Surface area= A 2 … n Length= L Fig. 3.5. Envisioning cement paste as serial layers I 34 In Equation 3.13 b is called the structure or the connectivity factor and represents the tortuosity (twistedness) of the layers. In a concrete matrix, conductivity of the saturated capillary porosity is several degrees of magnitude higher than the conductivity of other components. As a result, the total conductivity of the matrix is practically equal to the conductivity of the saturated capillary porosity: σ t ≈ σ oφ cap β (3.14) In this equation st is the overall conductivity of the matrix, so is the conductivity of pore solution, fcap is the volume fraction of capillary porosity, and b is a connectivity factor. Christensen (1993) showed that in a hydrating cement microstructure each of these parameters (so, fcap, and b) is changing with time and as a result, an overall change can be observed in the sample’s conductivity (Figure 3.6 (a) and (b)). 16 0 .6 14 10 0 .5 0 .4 8 0 .3 6 0 .2 1 s (s/m) 10 fcap so (s/m) 12 0.1 4 Series2 so (s/m) fcap Series1 2 0 .1 0 1 10 10 0 0 10 00 0.01 1 Age (hr) 10 100 1000 A ge (hr) (a) (b) Fig. 3.6. Variations in (a) pore solution conductivity (so) and volume fraction of capillary porosity (fcap) and (b) the complete sample’s conductivity (s) as a function of age (Christensen 1993) 35 While Christensen performed his tests on samples maintained at 100% relative humidity, McCarter and coworkers (1995 and 1996) studied conductivity variations in samples experiencing drying and rewetting. They used concrete slabs that were exposed to drying from one surface for a period of 16 weeks. After this period, the samples were ponded at the drying face with water or a 2 molar NaCl solution. Conductivity profiles of slabs were obtained before ponding (sB) and after 24 hours of ponding with water (sW) or NaCl solution (sN), (Figure 3.7). As it’s shown in the figure a significant increase in conductivity was observed after rewetting. The increase was more pronounced for NaCl ponded samples in comparison to water ponded samples due to dissolution of sodium and chloride ions in the pore solution and increasing its conductivity. . The research team proposed to link the ratio between conductivities before ponding and after water ponding (sB/sW) to the degree of saturation of concrete before rewetting via the following equation: σB = (S r ) m σW (3.15-a) when m is independent from porosity and moisture content of the sample and is in the region of 1.5 to 3.0 for cement pastes and in the range of 1.2 to 2.5 for concretes and mortars (McCarter et al. 1995). The research team also proposed to use the ratio between the conductivity of the water ponded sample to the conductivity of the NaCl ponded sample (sW/sN) to determine the concentration of NaCl dissolved in the pore solution: σ W σ oW = σ N σ oN (3.15-b) Using Equation 3.15-b, assuming that soW is known; soN can be calculated and be used to determine the concentration of chlorides in the pore solution of the NaCl ponded sample. 36 100 Bef ore ponding (B) Water ponding (W) NaCl ponding (N) 80 s (s/km) 60 40 20 0 0 10 20 30 40 50 Depth from ponding s urface (m m ) Fig. 3.7. The significant increase in the conductivity of dried concrete slabs due to rewetting with water and NaCl solution (McCarter et al. 1995) The current research intends to reframe and extend the previous findings in order to develop a method that accurately accounts for the humidity of concrete and to propose a procedure that is able to determine the internal humidity (or moisture content) of concrete based on the measured conductivity. For this purpose the relationships used by Christensen to relate conductivity to microstructural parameters were extended to also account for internal humidity of the microstructure. The procedure is as follows. Equation 3.14 is derived based on the assumption that all the capillary pores are saturated with pore solution. In a real life scenario a portion of pores are empty of liquid phase due to drying or self desiccation. In this case, only the portion of porosity that contains pore solution is conductive. To reflect this fact, the term fcap in Equation 3.14 should be substituted with fcap-con which represents the volume fraction of capillary porosity that is capable of conducting electrical current: σ t = σ oφ cap −con β (3.16) 37 When the system experiences a change in the internal moisture content, each of the three terms so, fcap-con, and b varies. For example, when a sample loses moisture, the concentration of ions in the pore solution increases which in turn causes an increase in the value of so. On the other hand, the volume of the liquid phase and consequently the volume of conductive porosity decreases (i.e., decrease in fcap-con). In addition, connectivity of pores changes which is reflected in a change in the value of b. Since simultaneous monitoring each of these parameters (so, fcap-con, and b) as a function of the sample’s internal humidity is difficult, in this study it is proposed that the conductivity changes caused by humidity variations are separated from the hydration effects through implementing a humidity factor (fH) in Equation 3.16. The procedure is to select a reference point at which the sample’s humidity is known and assign the humidity factor of “one” to this reference point (i.e., fH =1 at reference point). Then, the sample’s conductivity at other humidities is: σ t = (σ oφ cap −con β ) ref ⋅ f H (3.17) when fH is a function of the difference between the current humidity and the reference humidity: f H = f H (∆H = H ref − H cur ) (3.18) As an example a mature concrete can be considered in which changes in so, fcap-con, and b due to hydration is negligible. If one selects the 90% relative humidity as the reference point, conductivity of the sample at 85%RH can be obtained as follows: σ RH =85% = (σ oφ cap −con β ) RH =90% ⋅ f H ( ∆RH = +5%) (3.19) Figure 3.8 illustrates the idea. The humidity factor for a given microstructure can be obtained by measuring conductivity of the system at known humidities and establishing the humidity factor function (Equation 3.18). Then, when the internal humidity is unknown, measuring the microstructure’s conductivity can be used along with the obtained humidity factor function to back calculate the internal humidity (Equations 3.17 and 3.18). 38 reference state fH =1 less humidity fH <1 more humidity fH >1 Fig 3.8. The humidity factor represents the internal humidity of the system This is a valid approach as long as the microstructure is invariant with time. A cementitious microstructure experiences continuous hydrational developments and as a result the porosity and connectivity of the pores are changing as a function of time. Also, in a cementitious system the hydration rate and microstructural development are related to the microstructure’s humidity. However, for simplifying the process of calibration, in this study, the two following assumptions were made to enables application of the aforementioned approach to cementitious microstructures: 1- The microstructural development is independent of the moisture history and is only a function of age. 2- The humidity factor function for a given concrete mixture is independent of the specimens’ age. This means that one can obtain the “fH” function at a certain age and apply this function to the samples from the same mixture at different ages. Using these assumptions, the conductivity of two similar concrete samples (from the same mixture) cured at different relative humidities (for example one at 100%RH and the other at 85%RH) at all ages can be related via the following expression: σ RHcur = (σ oφ cap −con β ) RHref ⋅ f H ( ∆RH = RHref − RHcur ) (3.20) 39 As a result the function fH (DRH) can be obtained using a set of calibration specimens and be used to determine the humidity of similar concrete samples. The comprehensive details about obtaining and implementing the humidity factor function are described in Chapter 5. 3.4. Summary This chapter provided a review of the history and application of electrical measurements in concrete. The definition of electrical conductivity and the principles of direct (DC) and alternating (AC) conductivity measurements were presented. Conductivity of a cementitious microstructure is a function of the pore solution conductivity (so), the volume fraction of conductive porosity (fcap-con) and the connectivity of the pores (b). These parameters vary with the sample’s age and internal humidity. It is proposed in this study to differentiate between the changes caused by hydration and the changes caused by variations in internal humidity, through introducing a humidity factor to Equation 3.16. This approach can be used to determine the humidity of a sample based on its measured conductivity. approach are described in Chapter 5. The details of implementing this