On-site corrosion monitoring experience in concrete structures: potential improvements on the current-controlled polarization resistance method

2.1. Structures Inspected

Table 1 lists the most prominent characteristics of the structures assessed with MCC to determine their corrosion rate values. Most were inspected during the research team’s 13-year lifespan and a few earlier. The 15 structures involved are located in different areas of Spain and exposed to different aggressive environments, classified as per the categories set out in European standard EN 206-1:2000 (2626. EN 206:2013+A2:2021. Concrete-Part 1: Specification, performance, production and conformity. British Standards Institution, 2021.
). Although most were inspected on a one-off basis, some were assessed periodically over several years. A total of 1867 measurements were recorded, 891 or nearly half of which for a single structure, the first listed in Table 1. It and the following four structures together summed 75 % of the total readings, whilst the other 10 structures accounted for less than 5 % each.

Inasmuch as non-destructive techniques form part of more comprehensive inspections conducted on the structures studied, in some cases data were also available on concrete electrical resistivity. The Newman method (2727. Newman, J. (1966) Resistance for flow of current to a disk. J. Electrochem. Soc. 113 [5], 501-502. https://doi.org/10.1149/1.2424003.
) for calculating resistivity for the flow of current to a disk was applied to the reinforced concrete structures studied (2828. Feliu, S.; Andrade, C.; González, J.A.; Alonso, C. (1996) A new method for in-situ measurement of electrical resistivity of reinforced concrete. Mat. Struct. 29 [6], 362-365. https://doi.org/10.1007/BF02486344.
). A sensor with a disk diameter (Ø_DISC) of 2 cm and a cell constant k=2 was used to convert electrical resistance (R_E) to resistivity (ρ), assuming ρ=k·Ø_DISC·R_E. In the cases at issue the total chloride concentration on the concrete surface and at the depth of the first rebar were also found, applying the potentiometer method described in Spanish standard UNE 112010:2011 (2929. UNE 112010:2011 Spanish Standard, Corrosion of concrete reinforcement steel. Chloride determination for in-service concrete, 2011.
).

2.2. Corrosion Meter Operating Principle

The corrosion rates (expressed as corrosion current density, i_CORR) of the rebar embedded in the structures listed in sub-section 2.1 were measured in situ with a corrosion meter using the patented modulated current confinement method (MCC) (77. Feliu, S.; Gonzalez, J.A.; Feliu, V.; Escudero, L.; Rodriguez, I.A.; Ausin, V.; et al. inventors; CSIC Consejo Superior Investigaciones Científicas; Geotec Cimientos; Geotecnia y Cimientos Geocisa Sa; Consejo Superior Investigacion, assignee. Electrochemical measuring-corrosion rate of reinforcement in concrete-using DC to obtain corrosion rate taking into account reinforcement area affected by electrical signal patent CA2042883-A; ES2024268-A; US5259944-A.
). The device used is designed to also measure corrosion potential (E_CORR) and electrical resistance of concrete (R_E). The applicability of this type of electrochemical techniques to study reinforcement corrosion is based on the material’s electrolytic behavior, for the moisture locked in its complex interconnected pore network serves as the aqueous medium through which aggressive external agents penetrate the concrete cover to the embedded steel.

The corrosion meter used was not the commercial device based on the patent, however, but a laboratory prototype designed to manually vary internal measurement parameters such as polarization time and amplitude of the current pulse applied (I_AP). The prototype also delivered the values of the parameters characteristic of the test as the readings were taken, enabling researchers to analyze the method exhaustively in each measurement.

As noted earlier, the corrosion rate measured by the instrument was based on the Stern-Geary linear polarization resistance (R_P) method. The MCC method was used to control the area of reinforcing steel polarized with the galvanostatic pulse technique (GPT). The system entailed positioning a sensor equipped guard ring on the specimen to confine the signal applied to a controlled length of the reinforcement. R_P is obtained from the potential-time (E-t) response when the reinforcement was polarized with a galvanostatic pulse characterized by amplitude I_AP. The GPT deployed was designed around the equivalent circuit depicted in Figure 1.a proposed by Randles (3030. Feliu, V.; González, J.A.; Andrade, C.; Feliu, S. (1998) Equivalent circuit for modelling the steel-concrete interface. I. Experimental evidence and theoretical predictions. Corros. Sci. 40 [6], 975-993. https://doi.org/10.1016/S0010938X(98)00036-5.
) to model steel-concrete systems. Figure 1.b illustrates the potential response typical of that model. To find the i_CORR value, parameter R_P in Equation [1] was substituted for its numerical value as found with Equation [2]:

where E_POL is effective system polarization, obtained by subtracting the ohmic potential drop (E_Ω) from maximum polarization (E_MAX). The ohmic drop (E_Ω) appears as an abrupt variation in potential at the start and end (after interrupting the current) of the E-t curve. Its value is the product of I_AP·R_E, where R_E, concrete electrical resistance, is found from Equation [3]:

Given that GPT is based on the Stern-Geary method, the E_POL value must lie within 5 mV and 30 mV for corrosion rate calculations to be deemed correct. And since further to Equation [2], E_POL = R_P·I_AP; the I_AP value must be adjusted to corrosion kinetics (RP) to obtain a suitable EPOL. The corrosion meter therefore estimates the current to be applied (I_AP) for E_POL to lie within the desired range, although it is designed to ensure the I_AP is never under 5 µA. Pulse duration is an equally important factor, for it must be long enough to reach quasi-stationary status while the reading is being taken (E_MAX≈E_∞). Here the pulse was applied for 30 s to 100 s, depending on the characteristics of the steel-concrete system a hand. Pulses of around 30 s were used for corroded structures and close to 100 s for passive steel¹In each case, the decision regarding the polarization time is made by the researcher. The criterion is established by the qualitative value of the corrosion potential (E_CORR) measured prior to determining the corrosion rate. . Although I_AP was initially set to attempt to optimize E_POL, the value recorded and used to calculate R_P (Equation [2]) was the actual post-pulse value, i.e., the E_POL reading recorded when the current was interrupted (Figure 1.b).

Figure 2 shows the main components of the corrosion meter used for on-site corrosion rate measurements. The electronic element recorded reinforcement corrosion potential (E_CORR) and applied the current pulse, I_AP, with a galvanostat. That entailed connecting the instrument to the reinforcement and securing a guard ring to the surface of the structure along the length of rebar to be assessed (Figure 2.a).

The corrosion meter sensor (Figure 2.b) was fitted with a stainless-steel ring-shaped counter electrode (CE) through which the I_AP current was applied and a Cu/Cu₂SO₄ reference electrode (RE) in the centre to record the E-t response of the rebar, which is considered the working electrode (WE). The sensor also featured a sensorised guard ring to delimit the area of steel polarized (parameter A) to calculate i_CORR from Equation [1]. The guard ring consisted in an external counter electrode (XCE), also a ring-shaped stainless-steel component, through which current I_XCE is applied opposing that applied through the CE (I_CE = I_AP). The I_XCE value was adjusted to ensure the electrical field was confined to the area between the two rings on the grounds of the potential recorded by the two Cu/Cu₂SO₄ reference electrodes positioned between them (Figure 2.b). More specifically, the potential between the two electrodes had to be held at the same value as recorded prior to polarization.

The polarized reinforcement’s response to the galvanostatic pulse was received by the electronic element in the corrosion meter, which then processed the information to calculate corrosion parameters E_CORR, i_CORR and R_E. The values were displayed on the device screen and stored in its memory.

Parameter i_CORR was found by entering the R_P value recorded into Equation [1] and assuming a value of 26 mV for constant B in all cases. To define the working area (A) the sensor was assumed to project a 10.5 cm diameter cylinder inward toward the rebar. The length along the latter intersected by the projected cylinder was deemed to be the area where polarization was effective (Figure 2.a). In other words, the A value for each measurement depended on rebar diameter (Ø) and was calculated as:

Rebar arrangement in each of the structures listed in sub-section 2.1 was staked out prior to the first inspection. Corrosion meter readings were taken at the points defined to be the most representative or of greatest interest to the study underway at the time of inspection. Those points were identified after fully mapping E_CORR and electrical resistivity, which furnished qualitative information on the steel condition.

The corrosion meter was connected to the rebar to be assessed in each measurement and connected as well to the sensor on the instrument used to generate polarization (Figure 2). The sensor in turn was positioned on the concrete surface ensuring that the projected centreline of the rebar crossed through the centre of its circumference (Figure 2.b). Moist sponges were placed between the sensor electrodes and the concrete surface to facilitate the electrolytic connection between the measuring electrodes and the pore network in the concrete, which constitute the measuring electrolyte (Figure 2.a).

2.3. Criteria for Interpreting Measurements

The reliability of the measurements delivered by the corrosion meter described in sub-section 2.2 was verified by analyzing certain test parameters in two distinct stages of the process:

The criteria applied to the i_CORR findings to determine degree of reinforcement corrosion were as specified in Spanish standard UNE 112072:2011 (3131. UNE 112072:2011 Spanish Standard, Laboratory measurement of corrosion rate using the polarization resistance technique, 2011.
) and standards ASTM STP 1065 (3232. Andrade, C.; Alonso, M.C.; González, J.A. (1990) An initial effort to use the corrosion rate measurements for estimating rebar durability. In: Berke, N.S.; Chaker, V.; Whiting, D. editors. Corrosion rates of steel in concrete ASTM STP1065. Philadelphia: Am. Soc. Tes. Mat. 143.
) and RILEM TC-154-EMC (66. Andrade, C.; Alonso, C. (2004) Test methods for on-site corrosion rate measurement of steel reinforcement in concrete by means of the polarization resistance method. Mat. Struct. 37 [9], 623-643. https://doi.org/10.1007/BF02483292.
). Those findings were also compared to the values for other parameters of interest such as corrosion potential (E_CORR), further to the recommendations set out in (3333. ASTM C876 − 15, Standard Test Method for Corrosion Potentials of Uncoated Reinforcing Steel in Concrete. West Conshohocken, PA, 2015.
), and concrete resistivity (ρ) as laid down in (3434. Polder, R.B. (2001) Test methods for on site measurement of resistivity of concrete-A RILEM TC-154 technical recommendation. Constr. Build. Mater. 15 [2-3], 125-131. https://doi.org/10.1016/S0950-0618(00)00061-1.
).

3.1. Corrosion Meter Current Ratio

Figure 3 shows the percentage of measurements in which the ratio between currents I_CE and I_XCE could be deemed valid for the present purposes, with confinement satisfying the criteria described in sub-section 2.3. Outside of one extreme case (structure 2) where the confinement system was suitable for just 38.3% of measurements, 54% to 92% (mean = 74%) of the readings could be deemed valid on the grounds of the aforementioned inter-current ratio criteria and used for structural assessment and diagnosis.

Non-valid measurements were defined to be those where the corrosion meter guard ring failed to fully confine the pulse. At around 26%, the mean percentage of non-valid measurements might be a statistically assumable error, but in practice would mean readings would have to be taken by a qualified technician able to review each and repeat where necessary. That constraint may limit corrosion meter operability and lengthen structure inspection times.

Cases such as structure 2 may also arise, in which the non-valid rate was clearly lower than the mean. That development may have been related to the shortcomings reported by earlier authors in connection with the performance of ring guard corrosion meters in low electrical resistance (R_E) concretes (3535. Andrade, C.; Martínez, I.; Alonso, C.; Fullea, J. (2001) New advanced electrochemical techniques/or on site measurements of reinforcement corrosion. Mat. Constr. 51 [263-264], 97-107. https://doi.org/10.3989/mc.2001.v51.i263-264.356.
, 3636. Martínez, I.; Castillo, A. (2020) Concrete surface applied corrosion inhibitors: on site evaluation by non-destructive electrochemical techniques. In REHABEND 2020 Congress Granada (Spain), 24-27 March, 2020. https://hdl.handle.net/10261/234898.
), such as found in structure 2 in this study, exposed to a high chloride (XS3) environment.

The I_XCE/I_CE ratio is plotted against concrete electrical resistance (R_E) in Figure 4.a to study that issue in greater depth. The graph shows that as a rule the lower the R_E the more acute was the tendency for measurements to converge on I_XCE/I_CE ≈ 1, the limit below which the measurement was deemed to be non-valid due to corrosion meter guard ring malfunction. The rate of valid values would therefore be expected to be lower in the presence of low R_E values. And that is exactly the pattern observed in Figure 4.b, which graphs the percentage of valid measurements (I_XCE > I_CE) in the entire dataset for different ranges of R_E. The rate was particularly high (93.9% to 100%) for R_E values of over 10 kΩ, slid slightly for values of 1 kΩ to 10 kΩ (83.5%) and declined drastically to just 40% at R_E < 1 kΩ.

Of the three environment XS3 structures (2, 7 and 11) studied, structure 2 had the visibly lowest R_E values, most <1 kΩ (Figure 4.a), which would explain the high rate of non-valid measurements in that structure. The pattern observed in parameter R_E may be associated with concrete resistivity and chloride concentration, given as wt% (cement) in Table 2 for some of the structures assessed. Structures 2 and 11, exposed to XS3 (high chloride) environments, exhibited resistivities within the same order of magnitude (8.9 kΩ·cm and 5.2 kΩ·cm) and clearly lower than the others. That finding was related to both the degree of structural saturation and the chloride content in the concrete. Although the total surface chloride concentration was similar in those two structures (2.94% in 2 and 2.80% in 11), the total chloride content around the reinforcement, at 2.60%, was substantially higher in 2 than the 1.12% recorded in 11.

The foregoing discussion reveals a need to improve the guard ring system used in MCC method, for while its efficacy is acceptable as a rule, the rate of non-valid measurements delivered may be highly unacceptable in low R_E steel-concrete systems.

3.2. Reinforcement Polarization Potential

Figure 5.a graphs measurement reliability with respect to effective reinforcement polarization, i.e., polarization potential (E_POL), in keeping with the criteria described in sub-section 2.3, excluding the readings disregarded due to current imbalance in the corrosion meter.

That analysis also excluded structures where the number of measurements deemed as valid in terms of the E_POL value accounted for less than 3% of the total, for individualized analysis with so few data would not be reliable. In the eight structures ultimately analyzed, 32.6% to 66.7% (mean = 53.7%) of the measurements had an E_POL of 5 mV to 30 mV, the range compatible with reliable application of linear polarization resistance. In other words, approximately one of every two measurements was reliable. With such an unacceptably high error rate, valid use of the on-site measuring instrument would be subject to reviewing the E_POL value for each and every measurement to determine its accuracy and repeat the reading where necessary. That task, not always readily performed in situ, would need to be supplemented with the use of an ‘open’ device (such as the one adapted for this study) in which all the test parameters are visible, a feature missing in market instruments.

The measurements not reaching effective polarization values suitable for application of the linear polarization resistance method weighed in at 46.3% of the overall mean. Over-polarization (E_POL > 30 mV) prevailed, accounting for 31.6% compared to the 14.7% attributable to under-polarization (E_POL < 5 mV). That pattern differed from structure to structure when analyzed individually, however. The two types of failure were present to the same extent in structure 8, whilst measurements with E_POL < 5 mV prevailed in structures 5, 7 and especially 2. Measurements with E_POL > 30 mV predominated in structures 1, 3 4 and 6, with structures 3 and 6 exhibiting just 32.6% and 36% valid readings, respectively.

Figure 6, in turn, depicts the inter-relationships between E_POL, I_CE and corrosion density (i_CORR) for all eight structures addressed in this sub-section for a more exhaustive analysis of the type of failure (over- or under-polarization). Further to Figure 6.a, E_POL and i_CORR were proportionally and indirectly correlated. For E_POL values lying within the 5 mV to 30 mV interval, iCORR ranged from 0.01 µA/cm2 to 10 µA/cm² when the I_CE value was raised from the minimum device setting of 5 µA to a maximum of 80 µA. To better visualize this relationship, Figure 6.b graphs the percentage of measurements with E_POL < 5 mV, E_POL > 30 mV and E_POL = 5-30 mV for four ranges of i_CORR values.

The structures with passive reinforcement, i.e., where i_CORR was < 0.1 µA/cm², exhibited the lowest proportion of measurements with valid E_POL (5 mV to 30 mV), 30%. The percentage of valid measurements was highest, at 83.9%, for low corrosion (i_CORR from 0.1 µA/cm² to 0.5 µA/cm²), declined to 71.5% in the presence of moderate corrosion (0.5 µA/cm² to 1 µA/cm²) and to 57.5% where corrosion was high (> 1 µA/cm²). Where corrosion was moderate or high, the non-valid measurements (28.5% and 41.9%, respectively) were attributable exclusively to under-polarization (< 5 mV) (Figure 6.b). As Figure 6.a shows, such values were obtained because the I_CE applied proved to be insufficient to raise reinforcement polarization to above the desired threshold. In contrast, in passive reinforcement, where i_CORR < 0.1 µA/cm², the error (≈ 70%) was due exclusively to over-polarization (> 30 mV) (Figure 6.b). Although the minimum corrosion meter setting, I_CE = 5 µA (Figure 6.a) was applied in such cases, even that value proved to be too high.

Given those findings, the differences between structures identified in Figure 5.a can be analyzed against the percentage of valid measurements lying within each range of i_CORR values graphed in Figure 5.b. In structures 1, 3, 4 and 6, where the highest percentage of non-valid measurements was observed for E_POL > 30 mV (Figure 5.a), negligible i_CORR values (< 0.1 µA/cm²) prevailed. That effect was particularly prominent in structure 3, where 97.8% of the measurements showed an i_CORR of < 0.1 µA/cm², and would explain the 65.2% of non-valid readings attributable to over-polarization (E_POL > 30 mV). In structures 2, 5 and 7, in turn, non-valid measurements were primarily due to E_POL < 5 mV (Figure 5.a). The values observed for structure 2 were particularly striking, for nearly all its non-valid readings could be so attributed, a finding by no means unexpected inasmuch as 81.2% of its total measurements had an i_CORR of > 1 µA/cm². As noted in sub-section 3.1, structure 2 was exposed to an XS3 environment with a substantial chloride content, which would explain the high degree of corrosion recorded.

The inference to be drawn from the foregoing is that the MCC method must be improved by optimizing the I_CE value applied to ensure the polarization (E_POL) value remains within the acceptable range in the presence of both active and passive reinforcement.