Electrical impedance as a technique for civil engineer structures
surveillance. Considerations on the galvanic insulation of samples.
J M Torrents1, P Juan-García1 and A Aguado de Cea2
Technical University of Catalonia,
1
Department of Electronic Engineering. Jordi Girona 1-3, building C4, 08034 Barcelona (Spain)
2
Department of Construction Engineering. Jordi Girona 1-3, building C1, 08034 Barcelona (Spain)
E-mail: [email protected]
Abstract. Since Electric Impedance Spectroscopy (EIS) has been widely used to determine physical
properties of materials, it becomes necessary to evaluate different error contributions. In this work, it is
studied the effect of the current leakage due to the lack of galvanic insulation from sample to ground, which
could distort the results. In order to known the effects of ground coupling, an electric equivalent model is
developed to distinguish between the contribution of the sample impedance and the stray impedance one.
Model values agree with measured ones.
Keywords: Electrical impedance spectroscopy; galvanic insulation; ceramic materials.
1. Introduction
A Century ago, Schlumberger and Wenner among others in geoelectrical prospecting field located materials with
commercial interest under earth (sub-soil) thanks to their differential electrical characteristics from soil. Half
Century ago, Calleja (1952) studied electrical characteristics of cement and concretes in order to achieve better
knowledge of other physical characteristics of these ceramic materials. Later, some authors studied in depth
related subjects; as a few examples: (McCarter et al 1985, 1999), (Christensen et al 1994), (Xie et al 1994) or
(Yoon et al 1996). These authors demonstrated good correlation between electrical and structural or mechanical
characteristics of civil engineer materials.
In general, an electrical parameter is correlated with a mechanical parameter of interest. Often, the electrical
characteristic measured is Electrical Impedance (EI), Electrical Impedance Spectroscopy (EIS) or its directly
related parameters such resistivity or conductivity, mainly represented with two kinds of graphics (Nyquist or
Bode). In some applications, Electrical Impedance Tomography (EIT) is drawn for image mapping. Sometimes,
another electrical characteristic is measured (current, charge, voltage). Monitoring physical or mechanical
parameters of structures or materials through electrical parameters has plenty of advantages: electrical measures
are non destructive, usually easy to apply, with no risk and cheap (at least compared with other monitoring
alternatives).
Today, some authors have measured with better accuracy, other authors have measured impedance changes during
load or mechanical stress, others have analyzed (through impedance) components such fibers, or optimum SP
dose, or even corrosion effects, etc. As a few examples: Christensen et al (1992), Gu et al (1995), McCarter et al
(2004) have studied silica fume or fly ash behavior with electrical measurements. Andrade et al (1999) have
studied hardening process through EI. Torrents et al (1998) have compared different superplasticizer dose effect
on setting with a standard method (Vicat needle) and with conductivity. Cao and Chung (2001), Sun et al (2002)
have compared different load situations in structures with changes in their electrical parameters. Even more, Cao
et Chung (2001), Chung (2004) and Song (2006) have reported a kind of smartness in civil structures thanks to
any electrical parameter or actuation.
However, at the same time or even a little bit earlier, around the nineties, other authors have reported a few
limitations and difficulties in measuring with EI methods (Bari 1990), (Xie et al 1996), (Hsieh et al 1997),
(Hwang et al 1997), (Edwards et al 1997), (Mason et al 1998). These reported limitations and others not reported
yet probably are beneath the cause of all this success in lab results hardly transcend to field applications and
economical savings.
In this work, we study the current leakage effect due to the lack of galvanic insulation from sample to earth as one
of these undesired effects or limitations in materials characterization. In particular, we measure sample of
1
advanced cement based materials (concrete with steel fiber reinforcement). If an EIS study on a sample does not
take into account the current leakage effect, the results could be wrong. Our goal is provide a methodology for
detecting and avoiding the measurement problem of lack of insulation in order to make EIS or related techniques
more useful in field applications of material characterization.
2. Material and Methods
2.1. Sample preparation
Concrete prismatic samples (600x150x150 mm) were produced using CEM I 42.5 R Portland Cement. 2 types of
calcareous sand used (passing 2 and 4 mm sieve) was mixed with a sand/cement ratio of 3.4. The water/cement
ratio was 0.57. Aggregates were mixture gravel (passing 12 and 20 mm sieve). Polyfunctional additive was also
added.
50 mm long and 1.05 mm diameter Twinplate steel hooked end fibers were randomly mixed in different
proportions depending on the series measured. To study the electrical impedance dependence with the different
fiber loadings, series of 0, 20, 30, 40, 50 and 70 kg/m3 of fibers were tested.
The same steel fibers were used as electrodes for 4-wire measurements. Two electrodes were inserted at 5 and 10
cm from the sample edge (see figure 1) in the longitudinal axis of the sample and symmetrically to the transversal
axis were dipped the other two. Approximately half the fiber was inside the bulk so the other half was available to
connect each of the instrument terminals. Since it had to be placed into fresh paste, it allowed monitoring the bulk
impedance from the moment it started to harden. To avoid electrode oxidation, fibers had to be isolated as a
protection method and sanded down before measuring.
2.2. Measurement instruments
Electrical measurement parameters are usually performed with commercial instruments (Coombs 1995, p.27.7-
13), (Agilent 2006). To avoid the electrode impedance appearing in the final result, 4-wire measurements are
strongly recommended. 4-wire electrical impedance at low frequencies (below RF range) may be measured using
an auto-balanced bridge for its suitability in accuracy, resolution, ease to automate and, with newer instruments,
user friendly software.
We acquired data using a LF Impedance Analyzer HP4192A with a laboratory-made front-end based on a unit-
gain instrumentation amplifier that allows 4-wire impedance measurements developed from an idea published by
Gersing (1991). The Impedance Analyzer was controlled by a PC through a GPIB connection and software
developed under a LabWindows/CVI platform. In each prismatic sample, we measured 20 points per decade from
10 Hz to 10MHz with 1.1 Vrms voltage excitation. Figure 1 shows this setup schematically.
To measure capacitive coupling a Promax GF1000 function generator and a LeCroy 9314L oscilloscope were
used: we measured capacitance between all electrodes short-circuited and a metal platform (earth plane) where the
prismatic sample was placed.
Figure 1. Schematic diagram of the complete instrumentation system performing 4-wire electrical impedance
measurement on a 600x150x150 mm prismatic sample.
2
3. Results and discussion. Effects of earth coupling with prismatic samples
Concrete impedance, as many other similar composites, is mainly resistive at low frequencies, so a plateau
behavior in the low frequencies is expected (see figure 3a, cross (x) trace). This behavior contributes with an R
component in equivalent circuit models. A capacitive component, C, normally appears for frequencies over the
MHz range as well as other secondary effects (RC) related to the degree of porosity of the material (Cabeza et al,
2002). Besides, it is possible a tiny inductive effect (an L in equivalent circuit models) because current flow along
the sample and setup.
For safety reasons, the instrumentation equipment must to be electrically grounded [IEC 61140:2004]. A
capacitive coupling between ground and the prismatic sample arises giving an additional way to electrical current
to flow that is not detected by the measurement point in the 4-wire measurement. Coombs (1995, p. 27.16)
describes a typical (insulated) 4-point measurement. Therefore, it becomes an important error source due to the
difference between the injected and the detected current. Naturally, the bigger the sample is, the more important
effects of this error are.
Not only the size of the bulk affects to the final analyzer reading but also the relative position between the sample
and any earth point. Impedance measurement is supposed to be independent wherever the samples are placed, but
if there is an unknown current leakage to earth, it will be impossible to repeat any measurement in a different
place obtaining exactly the same results. Then, it is necessary to minimize the error placing the samples on an
insulated environment or to quantify the error effects at every frequency range in order to avoid them when
insulating is physically not possible.
Departing from other electrical models (Christensen et al, 1994), (Cabeza et al, 2002), it is essential to include in
the model the effects of ground coupling. A simple circuit with 3 branches in T-shape can include the effect of
ground coupling, where the two top shape parts model the bulk intrinsic impedance and the base T-shape part
models the feasible coupling between sample and earth (figure 2).
Figure 2. Impedance circuit based on RLC cell that models bulk impedance of samples and its undesired ground
coupling.
The influence of ground coupling was evaluated placing the sample on an earth plane and on an insulating
(timber) surface and comparing the results of both measures (with the same set up of figure 1). Since the sample is
physically in contact with the metallic plane, it can be considered an extreme coupling value. An unexpected peak
of impedance for frequencies from 10 kHz to 10 MHz is observed with its maximum around 1 MHz (figure 3a).
As described in section 2.2. Measurement instruments, capacitive coupling was quantified using the function
generator and the oscilloscope obtaining values around 2 nF for the peak frequencies. This obtained value will be
used to simulate an electrical model of the sample (see figure 3b).
To estimate the real values of the model components, the circuital model was adjusted to the measured data.
Figure 3a shows the agreement between simulated rhombuses (♦) trace and measured triangles (∆) trace values in
order to confirm the validity of the proposed model. It can be observed that the value of the stray capacitance is
the one manually measured previously. But even in that way, there is a slight discordance between both plots,
especially in the surroundings of the coupling peak. That occurs mainly because of the electrode contribution
combined with the earth coupling of the measured sample cannot be avoided (even with 4-wire measurements).
3
12000
Simulated Ground plane Insulated
10000
8000
Impedance moduli ( )
6000
4000
2000
0
10 100 1000 10000 100000 1000000 10000000
Frequency (Hz)
Figure 3. (a) Impedance moduli vs. frequency for sample measurements on ground plane triangles (∆) trace, on
an insulated environment crosses (x) trace and for simulated equivalent circuit rhombuses (♦) trace, and (b)
simulated circuit agreeing with measured data.
Theoretically, 4-point measurements should cancel the effects of electrode impedance but since the capacitive
ground coupling arises, it interferes with the auto-balanced bridge forming a voltage divider that causes a final
contribution of the electrode impedance in the final reading which not only depends on the own electrode
impedance but also on the ground coupling as well as on the intrinsic bulk impedance (1). This behavior
introduces a limitation to the Gersing (1991) proposal and shows that 4-wire measuring does not always guarantee
the absence of electrode contribution in the measurement. Circuital analysis of this phenomenon quantifies the
value of this limitation:
1 Z
ΔZ x = Z b e (1)
2 Zc
where ΔZx represents the erroneous increment of measured impedance due to earth coupling, Zb represents bulk
impedance, Ze models electrode impedance, and Zc represents ground coupling. In this particular setup, electrode
effects are remarkable for frequencies under 500 Hz and over 10 kHz, so an intermediate frequency is the
optimum to minimize the external errors. For instance, in the simulated measurement (see figure 3a), the
additional error due to the electrode contribution at 1 kHz is about 10 Ω, what represents 0.75% of relative error
(increased up to 7.5% at 10 kHz). Otherwise, if a wrong frequency is chosen, errors could exceed 100%,
becoming obviously unacceptable. In this experiment, it should be pointed out that measurements at 1 kHz
minimize the interfering errors, so can be compared between samples. However, since the dimension of the
interference depends on geometry, the optimum frequency must be carefully selected in each case.
4. Conclusions
A large number of techniques are based on measuring an electrical parameter with the purpose of studying another
mechanical one. Among others, impedance measurements are particularly common, as using non-destructive
techniques can becomes very useful for monitoring civil engineering structures as a technique for surveillance. An
electrical model allows the direct comparison between different specimens or geometries. However, when the
measured samples are big, a wide range of error sources should be evaluated, particularly those related with the
ground coupling. Since the impedance analyzer assumes that the current injected is exactly the same as the
detected, when an alternative current path arises, the detected impedance may be wrong. In order to avoid those
errors, is necessary to study its influence at each frequency ranges. Otherwise, erroneous measurements could
disguise the real values, and some disperse results be shown.
An equivalent circuit was obtained based on a parameter model, in order to simulate the effects of coupling
between the sample and the earth points of the setup. It is useful to compare measurements done over an earth
plane and on an isolated environment to quantify the stray capacitance but when measuring in field applications,
sometimes is not possible to achieve the necessary isolation so the practical methods consist on working on the
range of frequencies that minimize the error. The obtained electrical models help to understand the effects of
ground coupling and are effective to choose the best frequency to work in each situation. In the experiments,
4
measuring at 1 kHz was selected (with ground errors under 1%). When performing 4-wire measurements,
modeling ground coupling is not only necessary to evaluate the effects of current leakage but also to notice the
electrode contribution. Although it is supposed to be canceled by the 4-wire setup, ground coupling limits the
range of correct operation since the measurements are not completely independent of the electrodes impedance.
5. Acknowledgments
This work was partially supported by the Spanish Ministerio de Educación y Ciencia under project BIA 2005–
09412–C03–03 and it was also partially supported by the company FCCSA under project HATCONS.
6. References
Agilent Technologies; 2006. Impedance Measurement Handbook. Agilent Technologies.
Andrade C, Blanco VM, Collazo A, Keddam M, Novoa XR and Takenouti H, 1999. “Cement paste hardening
process studied by impedance spectroscopy”, Electrochimie Acta, 44 4313–8.
Bari M A, 1990. “Comment on 'Dynamic dielectric analysis during early-stage hydration of ordinary Portland
cement”, Journal Physics D: Applied Physics, 23 2 234-236.
Cabeza M, Merino P, Miranda A, Novoa X R and Sanchez I, 2002. “Impedance spectroscopy study of hardened
Portland cement paste”, Cement and Concrete Research, 32 881–91.
Calleja J, 1952. “New techniques in the study of setting and hardening of hydraulic materials”, Journal of the
American Concrete Institute, 23 525–36.
Cao J and Chung D D L, 2001. “Minor damage of cement mortar during cyclic compression monitored by
electrical resistivity measurements”, Cement and Concrete Research, 31 1519–21.
Christensen B J, Mason T O and Jennings H M, 1992. “Influence of silica fume on the early hydration of Portland
cements using impedance spectroscopy”, Journal of the American Ceramic Society, 75 4 939-945.
Christensen B J, Coverdale R T, Olson R A, Ford S J, Garboczi E J, Jennings H M and Mason T O, 1994.
“Impedance spectroscopy of hydrating cement based materials: measurement, interpretation, and application”,
Journal of the American Ceramic Society, 77 2789–804.
Chung D D L, 2004. “Self-heating structural materials”, Smart Material Structures, 13 3 562-565.
Coombs C F (ed), 1995. Electronic Instrument Handbook, 2nd ed. McGraw-Hill, New York. Chapter 27.
Edwards D D, Hwang J H, Ford S J and Mason T O, 1997. “Experimental limitations in impedance spectroscopy:
part V. Apparatus contributions and corrections”, Solid State Ionics, 99 1-2 85–93.
Gersing E, 1991. "Measurement of Electrical-Impedance in Organs - Measuring Equipment for Research and
Clinical-Applications." Biomedizinische Technik, 36 1-2 6-11.
Gu P, Xie P and Beaudoin J J, 1995. ”Determination of silica-fume content in hardened concrete by AC
impedance spectroscopy”, Cement and Concrete Research, 17 1 92-97.
Hsieh G, Ford S J, Mason T O and Pederson L R, 1997. “Experimental limitations in impedance spectroscopy.
Part VI. Four-point measurements of solid materials systems”, Solid State Ionics, 100 3-4 297-311.
Hwang J H, Kirkpatrick K S, Mason T O and Garboczi E J, 1997. “Experimental limitations in impedance
spectroscopy . Part IV. Electrode contact effects”, Solid State Ionics, 98 1-2 93-104.
IEC********
5
Mason T O, Ford S J, Shane J D, Hwang J H and Edwards D D, 1998. “Experimental limitations in impedance
spectroscopy of cement-based materials” Advances in Cement Research 10 4 143-150.
McCarter W J and Afshar A B, 1985. “Further studies on the early hydration of portland cement paste”. Journal
of Materials Science Letters, 4 4 405-408.
McCarter W J, Starrs G and Chrisp T M, 1999. “Immittance spectra for Portland cement/fly ash-based binders
during early hydration”, Cement and Concrete Research, 29 3 377-387.
McCarter W J, Starrs G and Chrisp T M., 2004. “The complex impedance response of fly-ash cements revisited”,
Cement and Concrete Research, 34 10 1837-1843.
Song G, Mo Y L, Otero K and Gu1 H, 2006. “Health monitoring and rehabilitation of a concrete structure using
intelligent materials”, Smart Material Structures, 15 309-314.
Torrents J M, Roncero J and Gettu R, 1998. “Utilization of impedance spectroscopy for studying the retarding
effect of a superplasticizer on the setting of cement”, Cement and Concrete Research, 28 9 1325-1333.
Xie P, Gu P, Fu Y and Beaudoin J J, 1994. “A.C. Impedance phenomena in hydrating cement systems: origin of
the high frequency ARC”, Cement and Concrete Research, 24 4 704-706.
Xie P, Gu P and Beaudoin J J, 1996. “Contact capacitance effect in measurement of a.c. impedance spectra for
hydrating cement systems”, Journal of Materials Science, 31 1 144-149.
Yoon S S, Kim H C and Hill R M, 1996. “The dielectric response of hydrating porous cement paste”, Journal of
Physics D: Applied Physics, 29 3 869-875.
