ABSTRACT
Cathodic protection (CP) of concrete structures has been successful in stopping corrosion in reinforcement for over 20 years in Europe. Long-term experience has been reported by various authors (Tinnea & Cryer, 2008; Nerland et al., 2007; Wenk & Oberhänsli, 2007; Polder, 1998). Advocates usually state that the service life of CP is superior to that of conventional repairs. Moreover, CP installations involve significant investment and their service life is an important issue. Service life aspects of CP have not received much attention, however. Analogous to modern service life design methods for concrete structures (e.g. DuraCrete), a performance based, limit state oriented and probabilistic approach should be chosen. This will allow CP to be incorporated into modern maintenance management systems, which surely will be performance oriented. This may be an unreachable ideal at the moment; however, this chapter intends to present some of the necessary elements. Such an approach is based on identifying the most important failure mechanisms, modelling their time-dependency, quantifying uncertainties and finally, calculating failure probabilities as a function of time. This chapter describes time-dependent degradation processes and failure mechanisms of essential components of impressed current CP systems in the operation stage. The starting point is that the CP system has been properly designed and executed and that commissioning, testing and voltage/current adjustment in the early stages have been carried out successfully.Good performance is defined as the absence of corrosion-induced damage. In practice, this is assumed to be the case if sufficient depolarisation is obtained (BS EN 12696). Sufficient depolarisation can only be reached when current flow is unhindered and control systems work properly. Normal maintenance should be applied, that is, two to four electrical checks and one visual inspection per year.Degradation of the anode material itself, loss of concrete/overlay adhesion, corrosion of anode-copper connections and failure of reference electrodes are the most significant time-dependent failure phenomena. Quantitative
models for degradation of the anode and overlay adhesion have been set up and validated from experimental evidence.As part of a study into CP service life, an inventory was made of systems operating in The Netherlands in 2004. The companies who installed them were asked about maintenance, performance and failure of components. Information was obtained from a total of 70 CP systems installed between 1987 and 2002. Some were not very well documented, however. Data on 52 of them provided sufficient information to be used. In particular the evaluation of anode-copper connection and reference electrode performance was based on the field data obtained. 6.2 Degradation of anode systems
CP current flow from the anode to the concrete involves oxidising and acidifying electrochemical reactions at the anode/concrete interface. Their type and relative proportion depend on the availability of reactants, the electrical potential and the anode material. Their rate depends mainly on the current density. At the anode/concrete (or anode/mortar) interface hydroxyl ions are oxidised to oxygen gas by: 4 OH-(aq) → O2 (g) + 2 H2O(I) + 4 e (6.1) This reaction tends to lower the pH and is equivalent to acid production. Hydroxyl ions from the hardened cement paste tend to buffer a high pH. Acid production may eventually dissolve the cement paste.When an oxidisable material such as carbon is present, it may be oxidised by: C(s) + 2 H2O(I) → CO2 (g) + 4 H+ (aq) + 4 e (6.2) or C(s) + H2O(I) → CO (g) + 2 H+ (aq) + 2 e (6.3) In addition to consuming carbon, these reactions also produce acid and tend to lower the pH. Reaction (6.2) occurs at a lower potential than reactions (6.1) or (6.3), so thermodynamically it should be favoured. Experiments in solution at high pH have shown that reaction (6.2) represents only about 20% of the current and reaction (6.1) consumes about 80% (Eastwood et al. 1999), so kinetics apparently dominate the course of the reactions. The slow kinetics of reactions (6.2) and (6.3) under many different conditions are well known from energy related research in which electrochemical oxidation of carbon (coal) was studied.Metals and chloride ions are potentially oxidisable. From the usual anode materials, titanium is strongly passivated (at normal potentials and pH) and the
‘activating’ noble metal oxides on the titanium surface cannot be oxidised any further. Oxidation of chloride is relatively small and can be neglected. 6.2.1 Oxidation of carbon-based materials
Carbon-based anode materials, such as carbon-filled conductive coatings, may suffer oxidising anodic attack (Brown & Tinnea; 1991). A tentative model can be based on Faraday’s law, taking the current density at the anode/concrete interface into account, which may differ from the current density per unit surface area of concrete. For a surface covering conductive coating, the anode/concrete surface ratio is 1. A carbon oxidation efficiency factor F(C) with a value between 0 and 1 is introduced, that takes into account that not all current oxidises carbon:
∆m(C) = F(C) * i(A) * A(carbon) * z-1 * F-1 (6.4) with ∆m(C) amount of carbon oxidised per unit of time [g/m2/s] F(C) the carbon oxidation efficiency factor [–], 0 < F(C) < 1 i(A) the anode/concrete surface area current density [A/m2] A(carbon) the atomic mass of carbon, 12 [g/mole] z the number of electrons involved in reaction (6.2), 4 [–] F Faraday’s constant, 96,500 [A s/mole]. For a surface covering carbon-based conductive coating CP system, a typical current density of 1 mA/m2 of concrete surface area equals an anode current density of 1 mA/m2. If the total current is oxidising carbon to carbon dioxide, so for F(C) = 1, this current density would oxidise 1.0 g of carbon per m2 of anode surface area per year.In more general terms, the amount of oxidation can be described with only one parameter, the oxidation efficiency factor F(C) and two input variables: the current density (assumed constant) and more conveniently written as I(A) in mA/m2 and time. Filling in numerical values for constants in (6.4) and multiplying by time in years, we obtain: ∆M(C) = 1.0 * F(C) * I(A) * t (6.5) with ∆M(C) the total amount of carbon oxidised per anode surface [g/m2] F(C) the carbon oxidation efficiency factor [–] I(A) the anode/concrete surface area current density [mA/m2] t the time [years]. For all the parameters listed above, time-averaged values should be used. Evidence from the literature to test this model is scarce. Observations on coating CP systems suggest a low level of attack (Polder et al., 2007). Samples from structures with a conductive coating (AHEAD from Protector)
that had undergone CP for up to 9 years at about 1 mA/m2 were studied by light microscopy (PFM) and scanning electron microscopy (SEM). The conductive coating had a thickness of about 150 µm, an estimated carbon content of 40% by volume (density 2250 kg/m3), totalling 135 g of carbon per m2 of concrete surface. If it had been oxidised at 1 mA/m2 for 9 years with F(C) = 1, about 9 g/m2 of carbon would have been oxidised, so ca. 7% of all carbon in the coating. Oxidation would have occurred near the coating/concrete interface first, then progressively deeper into the coating, as illustrated in Figure 6.1. After 9 years, a zone of at least 10 µm thick near the interface would have become relatively devoid of carbon particles. SEM would probably have shown this, which did not seem to be the case. This would also have increased the electrical resistance of the CP system, which was not reported. The evidence suggests that not all of the current is involved in oxidising carbon particles, so F(C) most likely had been significantly lower than 1.Another source does not relate to a conductive coating CP system, but may provide useful information. Mietz and colleagues have reported data on samples taken from a CP system in Berlin after 15 years (Mietz et al., 2001). This system used carbon filled polymer cables as anode (FEREX from Raychem) of 8 mm diameter with a 2 mm copper core. A length of 10.7 running metres of cable was used per m2 of concrete, so the anode/concrete surface ratio was about 0.25. During the first 7 years the system performed well, but it did not perform well during the remaining period. Increased system resistance required progressively increasing the driving voltage for sufficient protection (tested by depolarisation) in the later part of its life; at some point, sufficient depolarisation could not be reached any more. Autopsy of the anode cable was carried out using SEM and microprobe analysis after 15 years. It showed that carbon had disappeared from the outer layers of
Concrete
Conductive coating
New Aged
Carbon oxidation front moving in
the polymer cables, down to a maximum of about 2 mm depth. The total carbon content (supposedly graphite plus polymer) was reduced from 90% down to 74% in the degraded outer layers. These changes had occurred over depths of typically 0.5 – 1.0 mm and the mean current density was 2 mA/m2 (Mietz, personal communication, 2006). The amount of carbon oxidised in 15 years estimated from these analyses is equal to or greater than what corresponds to equation (6.5) with F(C) = 1. An explanation may be that the investigated samples had undergone a higher local current density than the average. Unfortunately, this renders these data unsuitable to obtain a value for F(C). However, this case led us to the conceptual model of a carbon oxidation front gradually moving into the carbon-filled anode material, resulting in increased resistance.Some other sources provide qualitative data. The performance of a Ferex system at Rotterdam (Schuten et al., 2007), was more favourable than that of the Berlin system. In Rotterdam, the anode system had not degraded significantly over 12 years of operation, as observed by light microscopy. Another Ferex system protecting 288 balconies at Capelle (NL) seemed to operate satisfactorily after 17 years (Nuiten, private communication, 2005). These cases may indicate that F(C) can be well below 1.A laboratory study (Eastwood et al., 1999) on carbon-filled polymer anodes (similar to Ferex) reported oxidation of carbon particles at high current efficiency in acid and neutral solutions (F(C) = 0.8 to 1). At high pH (> 11) and relatively high current densities, the carbon oxidation efficiency is low at about 20% and the remaining 80% of the current releases oxygen by reaction (6.1), so F(C) = 0.2. Most of the carbon seems to be oxidised to humic acids. Carbonate ions promote oxygen release and slow down carbon oxidation (at alkaline pH down to pH 9). Oxidation of carbon was found to increase the electrical resistance, very strongly at low pH and only moderately at high pH. Qualitatively, this corresponds well with the Berlin CP case. Taking into account that practical conditions may be more adverse than in the laboratory, we conclude that for the investigated anode material under normal CP operating conditions (a high pH, a relatively low anode potential and a low current density), F(C) may be about 0.3. Of course this value may differ for different carbon-based anode materials.Summarising, a conductive coating CP system can fail due to progressive oxidation of carbon particles from the anode/concrete interface. At some point in time, the electrical resistance of the coating layer that has become devoid of carbon particles is so high that sufficient current cannot be transferred to the concrete. Consequently, the system fails and the end of its working life has been reached. A similar type of failure has been reported for early carbon-based systems (Brown & Tinnea, 1991), particularly with high local current densities.
