The investigation of reinforced concrete structures that are suspected to be suffering from corrosion is commonly undertaken by specialist materials engi- neers working in conjunction with structural engineers. Before starting inspec- tion and testing, it is necessary to make a general assessment of the state of the structure concerned in order to confirm that it is reasonably safe to proceed with the proposed activity. It is also necessary to consider possible structural implications that may be associated with the various forms of corrosion (and/or other types of degradation) that may be identified. These aspects are within the province of the structural engineer and will not be discussed further in this contribution. Accurate diagnosis and location of the material degradation phenomena that have precipitated the investigation, and identification of their causative factors, however, are necessary precursors to any successful, cost- effective remedial treatment of the condition. Several useful accounts have been provided of the methodological steps and investigative techniques that are normally employed in carrying out such investigations (Pullar-Strecker, 1987; Broomfield, 1997; BRE, 2000b; Bertolini et al., 2004) and only a brief discussion of corrosion-related aspects will be given here. The initial requirements are usually to establish whether:
(a) characteristic symptoms of cracking and staining of the concrete, that might be attributable to corrosion, are present at the surface ± investigated by means of visual inspection and photography;
(b) there is evidence of sub-surface delamination, that might be attributable to corrosion ± usually investigated by means of a simple hammer/chain drag survey or, in some cases where large areas have to be surveyed, with the aid of instrumental methods, e.g. sub-surface radar or infra-red thermography;
(c) the distribution and cover depth of the reinforcing bars indicates that corrosion is likely to have been responsible for the above damage ±
investigated by means of an electromagnetic cover-meter survey;
(d) there is evidence of electrochemical corrosion activity, particularly in regions of the structure that have not already exhibited signs of cracking or delamination ± investigated by means of half-cell potential mapping (see Fig. 5.15), which is the subject of a recent draft RILEM Recommendation (RILEM TC 154-EMC, 2003).
Assuming the above indicators point to a corrosion problem, then the causative factors will need to be identified by measurements of the depths of carbonation and chloride analysis at representative locations. The phenolphtha- lein test, which is employed for measuring carbonation depths (RILEM TC CPC-18, 1985), is generally regarded as straightforward and reasonably reliable, though other methods are considered preferable in certain cases, e.g. when testing high alumina cement (HAC) concrete (BRE, 2000b). Chloride analyses, however, have sometimes proved problematic and large scatter in the results from round robin tests has been reported in various regions; these issues have been addressed by RILEM Technical Committee 178-TMC in recent reports and recommendations (Castellote and Andrade, 2001a,b). The question as to how data regarding carbonation depths and chloride profiles may be used to estimate corrosion initiation times was considered briefly in Sections 5.4±5.6 and, as already noted, particularly as far as chloride profiles are concerned, there are still a number of current uncertainties that remain to be resolved.
In cases where corrosion initiation is found already to have occurred and the causes have been identified, the issue of how to manage the problem for the remainder of the intended working life of the structure concerned raises questions of how rapidly the corrosion may be expected to propagate to a limit state at various locations. While potential mapping can provide a valuable guide to the distribution and scale of the anodic areas at a given time, it does not yield information about corrosion rates directly. Concrete resistivity measurements, which can be made non-destructively on site (Polder et al., 2000), may be correlated empirically with observed corrosion rates of steel after depassivation has occurred in concrete of particular composition but the relationship is not of universal applicability (Bertolini et al., 2004).
Attempts to estimate the instantaneous corrosion rates (icorr) of steel reinforcement in concrete directly may be made by various electrochemical techniques (Andrade and Alonso, 1996; Gowers and Millard, 1999), the most widely used of which involves adaptation of the well-known method of polarisation resistance (Rp) measurement (also known as linear polarisation) devised by Stern and coworkers (Stern and Geary, 1957, Stern and Weisert, 1958). The application of this method to measure the corrosion rate of steel in concrete was originally proposed and subsequently developed for on-site use by Andrade and co-workers (Andrade and Gonzalez, 1978; Feliu et al., 1989) and it has recently become the basis of a RILEM Recommendation (RILEM, 2004).
Although linear polarisation has been extensively applied to provide useful data about rates of corrosion of steel in concrete, it is recognised that there are inherent sources of error that limit the accuracy of the measurements, when compared with gravimetric estimates of corrosion (Andrade and Martinez, 2005). Particular problems arise from: (i) uncertainty in the magnitude of the proportionality constant (B) in the Stern-Geary equation, Rp B=icorr, (relating Rp to the reciprocal of the instantaneous corrosion rate, icorr) which varies for different systems, and (ii) errors in confining the areas of reinforcing steel undergoing polarisation when the technique is applied to a large reinforced concrete structure, as distinct from a small laboratory specimen. It is also important to note that instantaneous corrosion rates can vary substantially with time and ambient conditions, and so may not provide a representative guide to cumulative corrosion losses unless measurements are made repeatedly over an extended period at a given location.
For the above reasons, the use of icorr data in structural models aimed at providing quantitative predictions of the time for corrosion propagation to cause a limit state to be reached is subject to fairly large uncertainties. Commonly it is therefore assumed that, since a value of icorr 1 uAcm^-2 corresponds to a rate of loss of steel section of just over 10 um/year whilst somewhere between 10 and 100 um of steel section loss normally generates sufficient expansive corro- sion product to crack concrete cover (Broomfield, 1997), the broad criteria shown in Table 5.3 may be applied in the interpretation of icorr data for reinforced concrete.
While the various techniques described above can provide a means of assessing the corrosion state of an existing structure at a given time, it is becoming increasingly common for major structures in aggressive environments to have installed in them corrosion monitoring devices (sensors) that may be used to provide early warning of the development of corrosion. These may be based on a number of methods for non-destructive detection of changes that are associated with the onset of corrosive conditions. Examples include linear polarisation probes (Broomfield et al., 2002), galvanic sensors (Short et al., 1991, 1994), macrocell probes (Raupach and Schiessl, 1997), chloride ion specific electrodes and resistivity sensors (Zimmermann et al., 1997).